qemu-img.c 128.3 KB
Newer Older
B
bellard 已提交
1
/*
B
bellard 已提交
2
 * QEMU disk image utility
3
 *
B
bellard 已提交
4
 * Copyright (c) 2003-2008 Fabrice Bellard
5
 *
B
bellard 已提交
6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23
 * Permission is hereby granted, free of charge, to any person obtaining a copy
 * of this software and associated documentation files (the "Software"), to deal
 * in the Software without restriction, including without limitation the rights
 * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
 * copies of the Software, and to permit persons to whom the Software is
 * furnished to do so, subject to the following conditions:
 *
 * The above copyright notice and this permission notice shall be included in
 * all copies or substantial portions of the Software.
 *
 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
 * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
 * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
 * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
 * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
 * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
 * THE SOFTWARE.
 */
P
Peter Maydell 已提交
24
#include "qemu/osdep.h"
25
#include "qemu-version.h"
26
#include "qapi/error.h"
27
#include "qapi-visit.h"
28
#include "qapi/qobject-output-visitor.h"
29
#include "qapi/qmp/qerror.h"
30
#include "qapi/qmp/qjson.h"
31
#include "qapi/qmp/qbool.h"
32
#include "qemu/cutils.h"
33
#include "qemu/config-file.h"
34 35
#include "qemu/option.h"
#include "qemu/error-report.h"
36
#include "qemu/log.h"
37
#include "qom/object_interfaces.h"
38
#include "sysemu/sysemu.h"
M
Markus Armbruster 已提交
39
#include "sysemu/block-backend.h"
40
#include "block/block_int.h"
M
Max Reitz 已提交
41
#include "block/blockjob.h"
42
#include "block/qapi.h"
43
#include "crypto/init.h"
44
#include "trace/control.h"
45
#include <getopt.h>
46

47
#define QEMU_IMG_VERSION "qemu-img version " QEMU_VERSION QEMU_PKGVERSION \
48
                          "\n" QEMU_COPYRIGHT "\n"
49

A
Anthony Liguori 已提交
50
typedef struct img_cmd_t {
51 52
    const char *name;
    int (*handler)(int argc, char **argv);
A
Anthony Liguori 已提交
53
} img_cmd_t;
54

55 56 57
enum {
    OPTION_OUTPUT = 256,
    OPTION_BACKING_CHAIN = 257,
58
    OPTION_OBJECT = 258,
59
    OPTION_IMAGE_OPTS = 259,
K
Kevin Wolf 已提交
60
    OPTION_PATTERN = 260,
61 62
    OPTION_FLUSH_INTERVAL = 261,
    OPTION_NO_DRAIN = 262,
63 64 65 66 67 68 69
};

typedef enum OutputFormat {
    OFORMAT_JSON,
    OFORMAT_HUMAN,
} OutputFormat;

70
/* Default to cache=writeback as data integrity is not important for qemu-img */
71
#define BDRV_DEFAULT_CACHE "writeback"
72

73
static void format_print(void *opaque, const char *name)
B
bellard 已提交
74
{
75
    printf(" %s", name);
B
bellard 已提交
76 77
}

F
Fam Zheng 已提交
78 79 80 81 82 83 84 85 86 87 88 89 90 91
static void QEMU_NORETURN GCC_FMT_ATTR(1, 2) error_exit(const char *fmt, ...)
{
    va_list ap;

    error_printf("qemu-img: ");

    va_start(ap, fmt);
    error_vprintf(fmt, ap);
    va_end(ap);

    error_printf("\nTry 'qemu-img --help' for more information\n");
    exit(EXIT_FAILURE);
}

92 93 94 95 96 97 98 99 100 101
static void QEMU_NORETURN missing_argument(const char *option)
{
    error_exit("missing argument for option '%s'", option);
}

static void QEMU_NORETURN unrecognized_option(const char *option)
{
    error_exit("unrecognized option '%s'", option);
}

B
blueswir1 已提交
102
/* Please keep in synch with qemu-img.texi */
F
Fam Zheng 已提交
103
static void QEMU_NORETURN help(void)
B
bellard 已提交
104
{
105
    const char *help_msg =
106
           QEMU_IMG_VERSION
107
           "usage: qemu-img [standard options] command [command options]\n"
108 109
           "QEMU disk image utility\n"
           "\n"
110 111
           "    '-h', '--help'       display this help and exit\n"
           "    '-V', '--version'    output version information and exit\n"
112 113
           "    '-T', '--trace'      [[enable=]<pattern>][,events=<file>][,file=<file>]\n"
           "                         specify tracing options\n"
114
           "\n"
115
           "Command syntax:\n"
116 117 118 119 120
#define DEF(option, callback, arg_string)        \
           "  " arg_string "\n"
#include "qemu-img-cmds.h"
#undef DEF
#undef GEN_DOCS
121 122 123
           "\n"
           "Command parameters:\n"
           "  'filename' is a disk image filename\n"
124 125 126 127
           "  'objectdef' is a QEMU user creatable object definition. See the qemu(1)\n"
           "    manual page for a description of the object properties. The most common\n"
           "    object type is a 'secret', which is used to supply passwords and/or\n"
           "    encryption keys.\n"
128
           "  'fmt' is the disk image format. It is guessed automatically in most cases\n"
129
           "  'cache' is the cache mode used to write the output disk image, the valid\n"
130 131
           "    options are: 'none', 'writeback' (default, except for convert), 'writethrough',\n"
           "    'directsync' and 'unsafe' (default for convert)\n"
132 133
           "  'src_cache' is the cache mode used to read input disk images, the valid\n"
           "    options are the same as for the 'cache' option\n"
134
           "  'size' is the disk image size in bytes. Optional suffixes\n"
135 136 137
           "    'k' or 'K' (kilobyte, 1024), 'M' (megabyte, 1024k), 'G' (gigabyte, 1024M),\n"
           "    'T' (terabyte, 1024G), 'P' (petabyte, 1024T) and 'E' (exabyte, 1024P)  are\n"
           "    supported. 'b' is ignored.\n"
138 139 140 141 142
           "  'output_filename' is the destination disk image filename\n"
           "  'output_fmt' is the destination format\n"
           "  'options' is a comma separated list of format specific options in a\n"
           "    name=value format. Use -o ? for an overview of the options supported by the\n"
           "    used format\n"
143 144 145 146 147
           "  'snapshot_param' is param used for internal snapshot, format\n"
           "    is 'snapshot.id=[ID],snapshot.name=[NAME]', or\n"
           "    '[ID_OR_NAME]'\n"
           "  'snapshot_id_or_name' is deprecated, use 'snapshot_param'\n"
           "    instead\n"
148 149 150 151 152
           "  '-c' indicates that target image must be compressed (qcow format only)\n"
           "  '-u' enables unsafe rebasing. It is assumed that old and new backing file\n"
           "       match exactly. The image doesn't need a working backing file before\n"
           "       rebasing in this case (useful for renaming the backing file)\n"
           "  '-h' with or without a command shows this help and lists the supported formats\n"
153
           "  '-p' show progress of command (only certain commands)\n"
154
           "  '-q' use Quiet mode - do not print any output (except errors)\n"
155 156 157 158 159
           "  '-S' indicates the consecutive number of bytes (defaults to 4k) that must\n"
           "       contain only zeros for qemu-img to create a sparse image during\n"
           "       conversion. If the number of bytes is 0, the source will not be scanned for\n"
           "       unallocated or zero sectors, and the destination image will always be\n"
           "       fully allocated\n"
160
           "  '--output' takes the format in which the output must be done (human or json)\n"
161 162
           "  '-n' skips the target volume creation (useful if the volume is created\n"
           "       prior to running qemu-img)\n"
163
           "\n"
164 165 166 167
           "Parameters to check subcommand:\n"
           "  '-r' tries to repair any inconsistencies that are found during the check.\n"
           "       '-r leaks' repairs only cluster leaks, whereas '-r all' fixes all\n"
           "       kinds of errors, with a higher risk of choosing the wrong fix or\n"
168
           "       hiding corruption that has already occurred.\n"
169
           "\n"
P
Peter Lieven 已提交
170 171 172 173 174
           "Parameters to convert subcommand:\n"
           "  '-m' specifies how many coroutines work in parallel during the convert\n"
           "       process (defaults to 8)\n"
           "  '-W' allow to write to the target out of order rather than sequential\n"
           "\n"
175 176 177 178 179
           "Parameters to snapshot subcommand:\n"
           "  'snapshot' is the name of the snapshot to create, apply or delete\n"
           "  '-a' applies a snapshot (revert disk to saved state)\n"
           "  '-c' creates a snapshot\n"
           "  '-d' deletes a snapshot\n"
180 181 182 183 184
           "  '-l' lists all snapshots in the given image\n"
           "\n"
           "Parameters to compare subcommand:\n"
           "  '-f' first image format\n"
           "  '-F' second image format\n"
R
Reda Sallahi 已提交
185 186 187 188 189 190 191
           "  '-s' run in Strict mode - fail on different image size or sector allocation\n"
           "\n"
           "Parameters to dd subcommand:\n"
           "  'bs=BYTES' read and write up to BYTES bytes at a time "
           "(default: 512)\n"
           "  'count=N' copy only N input blocks\n"
           "  'if=FILE' read from FILE\n"
R
Reda Sallahi 已提交
192 193
           "  'of=FILE' write to FILE\n"
           "  'skip=N' skip N bs-sized blocks at the start of input\n";
194 195

    printf("%s\nSupported formats:", help_msg);
196
    bdrv_iterate_format(format_print, NULL);
B
bellard 已提交
197
    printf("\n");
F
Fam Zheng 已提交
198
    exit(EXIT_SUCCESS);
B
bellard 已提交
199 200
}

201 202 203 204 205 206 207 208 209
static QemuOptsList qemu_object_opts = {
    .name = "object",
    .implied_opt_name = "qom-type",
    .head = QTAILQ_HEAD_INITIALIZER(qemu_object_opts.head),
    .desc = {
        { }
    },
};

210 211 212 213 214 215 216 217 218
static QemuOptsList qemu_source_opts = {
    .name = "source",
    .implied_opt_name = "file",
    .head = QTAILQ_HEAD_INITIALIZER(qemu_source_opts.head),
    .desc = {
        { }
    },
};

S
Stefan Weil 已提交
219
static int GCC_FMT_ATTR(2, 3) qprintf(bool quiet, const char *fmt, ...)
220 221 222 223 224 225 226 227 228 229 230
{
    int ret = 0;
    if (!quiet) {
        va_list args;
        va_start(args, fmt);
        ret = vprintf(fmt, args);
        va_end(args);
    }
    return ret;
}

B
bellard 已提交
231

232 233 234
static int print_block_option_help(const char *filename, const char *fmt)
{
    BlockDriver *drv, *proto_drv;
235
    QemuOptsList *create_opts = NULL;
236
    Error *local_err = NULL;
237 238 239 240

    /* Find driver and parse its options */
    drv = bdrv_find_format(fmt);
    if (!drv) {
241
        error_report("Unknown file format '%s'", fmt);
242 243 244
        return 1;
    }

C
Chunyan Liu 已提交
245
    create_opts = qemu_opts_append(create_opts, drv->create_opts);
246
    if (filename) {
247
        proto_drv = bdrv_find_protocol(filename, true, &local_err);
248
        if (!proto_drv) {
249
            error_report_err(local_err);
250
            qemu_opts_free(create_opts);
251 252
            return 1;
        }
C
Chunyan Liu 已提交
253
        create_opts = qemu_opts_append(create_opts, proto_drv->create_opts);
254 255
    }

256 257
    qemu_opts_print_help(create_opts);
    qemu_opts_free(create_opts);
258 259 260
    return 0;
}

261 262

static int img_open_password(BlockBackend *blk, const char *filename,
263
                             int flags, bool quiet)
264 265 266
{
    BlockDriverState *bs;
    char password[256];
267 268

    bs = blk_bs(blk);
269 270
    if (bdrv_is_encrypted(bs) && bdrv_key_required(bs) &&
        !(flags & BDRV_O_NO_IO)) {
271 272 273 274 275 276 277 278 279 280 281 282 283 284
        qprintf(quiet, "Disk image '%s' is encrypted.\n", filename);
        if (qemu_read_password(password, sizeof(password)) < 0) {
            error_report("No password given");
            return -1;
        }
        if (bdrv_set_key(bs, password) < 0) {
            error_report("invalid password");
            return -1;
        }
    }
    return 0;
}


285
static BlockBackend *img_open_opts(const char *optstr,
286
                                   QemuOpts *opts, int flags, bool writethrough,
287
                                   bool quiet, bool force_share)
288 289 290 291 292
{
    QDict *options;
    Error *local_err = NULL;
    BlockBackend *blk;
    options = qemu_opts_to_qdict(opts, NULL);
293 294 295 296 297 298 299 300
    if (force_share) {
        if (qdict_haskey(options, BDRV_OPT_FORCE_SHARE)
            && !qdict_get_bool(options, BDRV_OPT_FORCE_SHARE)) {
            error_report("--force-share/-U conflicts with image options");
            return NULL;
        }
        qdict_put(options, BDRV_OPT_FORCE_SHARE, qbool_from_bool(true));
    }
301
    blk = blk_new_open(NULL, NULL, options, flags, &local_err);
302
    if (!blk) {
303
        error_reportf_err(local_err, "Could not open '%s': ", optstr);
304 305
        return NULL;
    }
306
    blk_set_enable_write_cache(blk, !writethrough);
307

308
    if (img_open_password(blk, optstr, flags, quiet) < 0) {
309 310 311 312 313 314
        blk_unref(blk);
        return NULL;
    }
    return blk;
}

315
static BlockBackend *img_open_file(const char *filename,
316
                                   const char *fmt, int flags,
317 318
                                   bool writethrough, bool quiet,
                                   bool force_share)
319 320
{
    BlockBackend *blk;
321
    Error *local_err = NULL;
322
    QDict *options = qdict_new();
323

324
    if (fmt) {
325
        qdict_put_str(options, "driver", fmt);
326
    }
327

328 329 330
    if (force_share) {
        qdict_put(options, BDRV_OPT_FORCE_SHARE, qbool_from_bool(true));
    }
331
    blk = blk_new_open(filename, NULL, options, flags, &local_err);
332
    if (!blk) {
333
        error_reportf_err(local_err, "Could not open '%s': ", filename);
334
        return NULL;
335
    }
336
    blk_set_enable_write_cache(blk, !writethrough);
337

338
    if (img_open_password(blk, filename, flags, quiet) < 0) {
339 340 341 342 343 344 345
        blk_unref(blk);
        return NULL;
    }
    return blk;
}


346
static BlockBackend *img_open(bool image_opts,
347
                              const char *filename,
348
                              const char *fmt, int flags, bool writethrough,
349
                              bool quiet, bool force_share)
350 351 352 353 354 355 356 357 358 359 360 361 362
{
    BlockBackend *blk;
    if (image_opts) {
        QemuOpts *opts;
        if (fmt) {
            error_report("--image-opts and --format are mutually exclusive");
            return NULL;
        }
        opts = qemu_opts_parse_noisily(qemu_find_opts("source"),
                                       filename, true);
        if (!opts) {
            return NULL;
        }
363 364
        blk = img_open_opts(filename, opts, flags, writethrough, quiet,
                            force_share);
365
    } else {
366 367
        blk = img_open_file(filename, fmt, flags, writethrough, quiet,
                            force_share);
368
    }
369
    return blk;
370 371
}

372

373
static int add_old_style_options(const char *fmt, QemuOpts *opts,
374 375
                                 const char *base_filename,
                                 const char *base_fmt)
376
{
377 378
    Error *err = NULL;

379
    if (base_filename) {
380
        qemu_opt_set(opts, BLOCK_OPT_BACKING_FILE, base_filename, &err);
381
        if (err) {
382 383
            error_report("Backing file not supported for file format '%s'",
                         fmt);
384
            error_free(err);
385
            return -1;
386 387 388
        }
    }
    if (base_fmt) {
389
        qemu_opt_set(opts, BLOCK_OPT_BACKING_FMT, base_fmt, &err);
390
        if (err) {
391 392
            error_report("Backing file format not supported for file "
                         "format '%s'", fmt);
393
            error_free(err);
394
            return -1;
395 396
        }
    }
397
    return 0;
398 399
}

400 401
static int64_t cvtnum(const char *s)
{
402
    int err;
403
    uint64_t value;
404

405 406 407 408
    err = qemu_strtosz(s, NULL, &value);
    if (err < 0) {
        return err;
    }
409 410 411
    if (value > INT64_MAX) {
        return -ERANGE;
    }
412
    return value;
413 414
}

B
bellard 已提交
415 416
static int img_create(int argc, char **argv)
{
417
    int c;
418
    uint64_t img_size = -1;
B
bellard 已提交
419
    const char *fmt = "raw";
420
    const char *base_fmt = NULL;
B
bellard 已提交
421 422
    const char *filename;
    const char *base_filename = NULL;
423
    char *options = NULL;
424
    Error *local_err = NULL;
425
    bool quiet = false;
426

B
bellard 已提交
427
    for(;;) {
428 429 430 431 432
        static const struct option long_options[] = {
            {"help", no_argument, 0, 'h'},
            {"object", required_argument, 0, OPTION_OBJECT},
            {0, 0, 0, 0}
        };
433
        c = getopt_long(argc, argv, ":F:b:f:he6o:q",
434
                        long_options, NULL);
435
        if (c == -1) {
B
bellard 已提交
436
            break;
437
        }
B
bellard 已提交
438
        switch(c) {
439 440 441
        case ':':
            missing_argument(argv[optind - 1]);
            break;
J
Jes Sorensen 已提交
442
        case '?':
443 444
            unrecognized_option(argv[optind - 1]);
            break;
B
bellard 已提交
445 446 447
        case 'h':
            help();
            break;
448 449 450
        case 'F':
            base_fmt = optarg;
            break;
B
bellard 已提交
451 452 453 454 455 456 457
        case 'b':
            base_filename = optarg;
            break;
        case 'f':
            fmt = optarg;
            break;
        case 'e':
458
            error_report("option -e is deprecated, please use \'-o "
459
                  "encryption\' instead!");
460
            goto fail;
461
        case '6':
462
            error_report("option -6 is deprecated, please use \'-o "
463
                  "compat6\' instead!");
464
            goto fail;
465
        case 'o':
466 467 468 469 470 471 472 473 474 475 476
            if (!is_valid_option_list(optarg)) {
                error_report("Invalid option list: %s", optarg);
                goto fail;
            }
            if (!options) {
                options = g_strdup(optarg);
            } else {
                char *old_options = options;
                options = g_strdup_printf("%s,%s", options, optarg);
                g_free(old_options);
            }
477
            break;
478 479 480
        case 'q':
            quiet = true;
            break;
481 482 483 484 485 486 487 488
        case OPTION_OBJECT: {
            QemuOpts *opts;
            opts = qemu_opts_parse_noisily(&qemu_object_opts,
                                           optarg, true);
            if (!opts) {
                goto fail;
            }
        }   break;
B
bellard 已提交
489 490
        }
    }
491

492
    /* Get the filename */
493 494 495 496 497 498
    filename = (optind < argc) ? argv[optind] : NULL;
    if (options && has_help_option(options)) {
        g_free(options);
        return print_block_option_help(filename, fmt);
    }

499
    if (optind >= argc) {
F
Fam Zheng 已提交
500
        error_exit("Expecting image file name");
501
    }
502
    optind++;
503

504 505
    if (qemu_opts_foreach(&qemu_object_opts,
                          user_creatable_add_opts_foreach,
506
                          NULL, NULL)) {
507 508 509
        goto fail;
    }

510 511
    /* Get image size, if specified */
    if (optind < argc) {
512
        int64_t sval;
513 514 515

        sval = cvtnum(argv[optind++]);
        if (sval < 0) {
516 517 518 519
            if (sval == -ERANGE) {
                error_report("Image size must be less than 8 EiB!");
            } else {
                error_report("Invalid image size specified! You may use k, M, "
520 521 522
                      "G, T, P or E suffixes for ");
                error_report("kilobytes, megabytes, gigabytes, terabytes, "
                             "petabytes and exabytes.");
523
            }
524
            goto fail;
525 526 527
        }
        img_size = (uint64_t)sval;
    }
528
    if (optind != argc) {
F
Fam Zheng 已提交
529
        error_exit("Unexpected argument: %s", argv[optind]);
530
    }
531

532
    bdrv_img_create(filename, fmt, base_filename, base_fmt,
533
                    options, img_size, 0, quiet, &local_err);
534
    if (local_err) {
535
        error_reportf_err(local_err, "%s: ", filename);
536
        goto fail;
537
    }
538

539
    g_free(options);
B
bellard 已提交
540
    return 0;
541 542 543 544

fail:
    g_free(options);
    return 1;
B
bellard 已提交
545 546
}

547
static void dump_json_image_check(ImageCheck *check, bool quiet)
548 549 550
{
    QString *str;
    QObject *obj;
551
    Visitor *v = qobject_output_visitor_new(&obj);
552 553 554

    visit_type_ImageCheck(v, NULL, &check, &error_abort);
    visit_complete(v, &obj);
555 556
    str = qobject_to_json_pretty(obj);
    assert(str != NULL);
557
    qprintf(quiet, "%s\n", qstring_get_str(str));
558
    qobject_decref(obj);
559
    visit_free(v);
560 561 562
    QDECREF(str);
}

563
static void dump_human_image_check(ImageCheck *check, bool quiet)
564 565
{
    if (!(check->corruptions || check->leaks || check->check_errors)) {
566
        qprintf(quiet, "No errors were found on the image.\n");
567 568
    } else {
        if (check->corruptions) {
569 570 571 572
            qprintf(quiet, "\n%" PRId64 " errors were found on the image.\n"
                    "Data may be corrupted, or further writes to the image "
                    "may corrupt it.\n",
                    check->corruptions);
573 574 575
        }

        if (check->leaks) {
576 577 578 579
            qprintf(quiet,
                    "\n%" PRId64 " leaked clusters were found on the image.\n"
                    "This means waste of disk space, but no harm to data.\n",
                    check->leaks);
580 581 582
        }

        if (check->check_errors) {
583 584 585 586
            qprintf(quiet,
                    "\n%" PRId64
                    " internal errors have occurred during the check.\n",
                    check->check_errors);
587 588 589 590
        }
    }

    if (check->total_clusters != 0 && check->allocated_clusters != 0) {
591 592 593 594 595 596 597
        qprintf(quiet, "%" PRId64 "/%" PRId64 " = %0.2f%% allocated, "
                "%0.2f%% fragmented, %0.2f%% compressed clusters\n",
                check->allocated_clusters, check->total_clusters,
                check->allocated_clusters * 100.0 / check->total_clusters,
                check->fragmented_clusters * 100.0 / check->allocated_clusters,
                check->compressed_clusters * 100.0 /
                check->allocated_clusters);
598 599 600
    }

    if (check->image_end_offset) {
601 602
        qprintf(quiet,
                "Image end offset: %" PRId64 "\n", check->image_end_offset);
603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638
    }
}

static int collect_image_check(BlockDriverState *bs,
                   ImageCheck *check,
                   const char *filename,
                   const char *fmt,
                   int fix)
{
    int ret;
    BdrvCheckResult result;

    ret = bdrv_check(bs, &result, fix);
    if (ret < 0) {
        return ret;
    }

    check->filename                 = g_strdup(filename);
    check->format                   = g_strdup(bdrv_get_format_name(bs));
    check->check_errors             = result.check_errors;
    check->corruptions              = result.corruptions;
    check->has_corruptions          = result.corruptions != 0;
    check->leaks                    = result.leaks;
    check->has_leaks                = result.leaks != 0;
    check->corruptions_fixed        = result.corruptions_fixed;
    check->has_corruptions_fixed    = result.corruptions != 0;
    check->leaks_fixed              = result.leaks_fixed;
    check->has_leaks_fixed          = result.leaks != 0;
    check->image_end_offset         = result.image_end_offset;
    check->has_image_end_offset     = result.image_end_offset != 0;
    check->total_clusters           = result.bfi.total_clusters;
    check->has_total_clusters       = result.bfi.total_clusters != 0;
    check->allocated_clusters       = result.bfi.allocated_clusters;
    check->has_allocated_clusters   = result.bfi.allocated_clusters != 0;
    check->fragmented_clusters      = result.bfi.fragmented_clusters;
    check->has_fragmented_clusters  = result.bfi.fragmented_clusters != 0;
639 640
    check->compressed_clusters      = result.bfi.compressed_clusters;
    check->has_compressed_clusters  = result.bfi.compressed_clusters != 0;
641 642 643 644

    return 0;
}

645 646 647
/*
 * Checks an image for consistency. Exit codes:
 *
M
Max Reitz 已提交
648 649 650 651 652
 *  0 - Check completed, image is good
 *  1 - Check not completed because of internal errors
 *  2 - Check completed, image is corrupted
 *  3 - Check completed, image has leaked clusters, but is good otherwise
 * 63 - Checks are not supported by the image format
653
 */
654 655 656
static int img_check(int argc, char **argv)
{
    int c, ret;
657
    OutputFormat output_format = OFORMAT_HUMAN;
658
    const char *filename, *fmt, *output, *cache;
M
Markus Armbruster 已提交
659
    BlockBackend *blk;
660
    BlockDriverState *bs;
661
    int fix = 0;
662 663
    int flags = BDRV_O_CHECK;
    bool writethrough;
664
    ImageCheck *check;
665
    bool quiet = false;
666
    bool image_opts = false;
667
    bool force_share = false;
668 669

    fmt = NULL;
670
    output = NULL;
671
    cache = BDRV_DEFAULT_CACHE;
672

673
    for(;;) {
674 675 676 677
        int option_index = 0;
        static const struct option long_options[] = {
            {"help", no_argument, 0, 'h'},
            {"format", required_argument, 0, 'f'},
678
            {"repair", required_argument, 0, 'r'},
679
            {"output", required_argument, 0, OPTION_OUTPUT},
680
            {"object", required_argument, 0, OPTION_OBJECT},
681
            {"image-opts", no_argument, 0, OPTION_IMAGE_OPTS},
682
            {"force-share", no_argument, 0, 'U'},
683 684
            {0, 0, 0, 0}
        };
685
        c = getopt_long(argc, argv, ":hf:r:T:qU",
686
                        long_options, &option_index);
687
        if (c == -1) {
688
            break;
689
        }
690
        switch(c) {
691 692 693
        case ':':
            missing_argument(argv[optind - 1]);
            break;
J
Jes Sorensen 已提交
694
        case '?':
695 696
            unrecognized_option(argv[optind - 1]);
            break;
697 698 699 700 701 702
        case 'h':
            help();
            break;
        case 'f':
            fmt = optarg;
            break;
703 704 705 706 707 708 709 710
        case 'r':
            flags |= BDRV_O_RDWR;

            if (!strcmp(optarg, "leaks")) {
                fix = BDRV_FIX_LEAKS;
            } else if (!strcmp(optarg, "all")) {
                fix = BDRV_FIX_LEAKS | BDRV_FIX_ERRORS;
            } else {
F
Fam Zheng 已提交
711 712
                error_exit("Unknown option value for -r "
                           "(expecting 'leaks' or 'all'): %s", optarg);
713 714
            }
            break;
715 716 717
        case OPTION_OUTPUT:
            output = optarg;
            break;
718 719 720
        case 'T':
            cache = optarg;
            break;
721 722 723
        case 'q':
            quiet = true;
            break;
724 725 726
        case 'U':
            force_share = true;
            break;
727 728 729 730 731 732 733 734
        case OPTION_OBJECT: {
            QemuOpts *opts;
            opts = qemu_opts_parse_noisily(&qemu_object_opts,
                                           optarg, true);
            if (!opts) {
                return 1;
            }
        }   break;
735 736 737
        case OPTION_IMAGE_OPTS:
            image_opts = true;
            break;
738 739
        }
    }
740
    if (optind != argc - 1) {
F
Fam Zheng 已提交
741
        error_exit("Expecting one image file name");
742
    }
743 744
    filename = argv[optind++];

745 746 747 748 749 750 751 752 753
    if (output && !strcmp(output, "json")) {
        output_format = OFORMAT_JSON;
    } else if (output && !strcmp(output, "human")) {
        output_format = OFORMAT_HUMAN;
    } else if (output) {
        error_report("--output must be used with human or json as argument.");
        return 1;
    }

754 755
    if (qemu_opts_foreach(&qemu_object_opts,
                          user_creatable_add_opts_foreach,
756
                          NULL, NULL)) {
757 758 759
        return 1;
    }

760
    ret = bdrv_parse_cache_mode(cache, &flags, &writethrough);
761 762 763 764 765
    if (ret < 0) {
        error_report("Invalid source cache option: %s", cache);
        return 1;
    }

766 767
    blk = img_open(image_opts, filename, fmt, flags, writethrough, quiet,
                   force_share);
768 769
    if (!blk) {
        return 1;
770
    }
771
    bs = blk_bs(blk);
772 773 774

    check = g_new0(ImageCheck, 1);
    ret = collect_image_check(bs, check, filename, fmt, fix);
775 776

    if (ret == -ENOTSUP) {
777
        error_report("This image format does not support checks");
778
        ret = 63;
779
        goto fail;
780 781
    }

782 783
    if (check->corruptions_fixed || check->leaks_fixed) {
        int corruptions_fixed, leaks_fixed;
784

785 786
        leaks_fixed         = check->leaks_fixed;
        corruptions_fixed   = check->corruptions_fixed;
787

788
        if (output_format == OFORMAT_HUMAN) {
789 790 791 792 793 794 795
            qprintf(quiet,
                    "The following inconsistencies were found and repaired:\n\n"
                    "    %" PRId64 " leaked clusters\n"
                    "    %" PRId64 " corruptions\n\n"
                    "Double checking the fixed image now...\n",
                    check->leaks_fixed,
                    check->corruptions_fixed);
796 797
        }

798
        ret = collect_image_check(bs, check, filename, fmt, 0);
799

800 801
        check->leaks_fixed          = leaks_fixed;
        check->corruptions_fixed    = corruptions_fixed;
802 803
    }

M
Max Reitz 已提交
804 805 806 807 808 809 810 811 812
    if (!ret) {
        switch (output_format) {
        case OFORMAT_HUMAN:
            dump_human_image_check(check, quiet);
            break;
        case OFORMAT_JSON:
            dump_json_image_check(check, quiet);
            break;
        }
813 814
    }

815
    if (ret || check->check_errors) {
M
Max Reitz 已提交
816 817 818 819 820
        if (ret) {
            error_report("Check failed: %s", strerror(-ret));
        } else {
            error_report("Check failed");
        }
821 822
        ret = 1;
        goto fail;
823
    }
824

825 826 827 828
    if (check->corruptions) {
        ret = 2;
    } else if (check->leaks) {
        ret = 3;
829
    } else {
830
        ret = 0;
831
    }
832 833 834

fail:
    qapi_free_ImageCheck(check);
M
Markus Armbruster 已提交
835
    blk_unref(blk);
836
    return ret;
837 838
}

M
Max Reitz 已提交
839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854
typedef struct CommonBlockJobCBInfo {
    BlockDriverState *bs;
    Error **errp;
} CommonBlockJobCBInfo;

static void common_block_job_cb(void *opaque, int ret)
{
    CommonBlockJobCBInfo *cbi = opaque;

    if (ret < 0) {
        error_setg_errno(cbi->errp, -ret, "Block job failed");
    }
}

static void run_block_job(BlockJob *job, Error **errp)
{
K
Kevin Wolf 已提交
855
    AioContext *aio_context = blk_get_aio_context(job->blk);
M
Max Reitz 已提交
856

857 858
    /* FIXME In error cases, the job simply goes away and we access a dangling
     * pointer below. */
859
    aio_context_acquire(aio_context);
M
Max Reitz 已提交
860 861
    do {
        aio_poll(aio_context, true);
862 863
        qemu_progress_print(job->len ?
                            ((float)job->offset / job->len * 100.f) : 0.0f, 0);
M
Max Reitz 已提交
864 865 866
    } while (!job->ready);

    block_job_complete_sync(job, errp);
867
    aio_context_release(aio_context);
868 869 870 871

    /* A block job may finish instantaneously without publishing any progress,
     * so just signal completion here */
    qemu_progress_print(100.f, 0);
M
Max Reitz 已提交
872 873
}

B
bellard 已提交
874 875
static int img_commit(int argc, char **argv)
{
876
    int c, ret, flags;
877
    const char *filename, *fmt, *cache, *base;
M
Markus Armbruster 已提交
878
    BlockBackend *blk;
M
Max Reitz 已提交
879
    BlockDriverState *bs, *base_bs;
880
    BlockJob *job;
881
    bool progress = false, quiet = false, drop = false;
882
    bool writethrough;
M
Max Reitz 已提交
883 884
    Error *local_err = NULL;
    CommonBlockJobCBInfo cbi;
885
    bool image_opts = false;
886
    AioContext *aio_context;
B
bellard 已提交
887 888

    fmt = NULL;
889
    cache = BDRV_DEFAULT_CACHE;
890
    base = NULL;
B
bellard 已提交
891
    for(;;) {
892 893 894
        static const struct option long_options[] = {
            {"help", no_argument, 0, 'h'},
            {"object", required_argument, 0, OPTION_OBJECT},
895
            {"image-opts", no_argument, 0, OPTION_IMAGE_OPTS},
896 897
            {0, 0, 0, 0}
        };
898
        c = getopt_long(argc, argv, ":f:ht:b:dpq",
899
                        long_options, NULL);
900
        if (c == -1) {
B
bellard 已提交
901
            break;
902
        }
B
bellard 已提交
903
        switch(c) {
904 905 906
        case ':':
            missing_argument(argv[optind - 1]);
            break;
J
Jes Sorensen 已提交
907
        case '?':
908 909
            unrecognized_option(argv[optind - 1]);
            break;
B
bellard 已提交
910 911 912 913 914 915
        case 'h':
            help();
            break;
        case 'f':
            fmt = optarg;
            break;
916 917 918
        case 't':
            cache = optarg;
            break;
919 920 921 922 923
        case 'b':
            base = optarg;
            /* -b implies -d */
            drop = true;
            break;
M
Max Reitz 已提交
924 925 926
        case 'd':
            drop = true;
            break;
927 928 929
        case 'p':
            progress = true;
            break;
930 931 932
        case 'q':
            quiet = true;
            break;
933 934 935 936 937 938 939 940
        case OPTION_OBJECT: {
            QemuOpts *opts;
            opts = qemu_opts_parse_noisily(&qemu_object_opts,
                                           optarg, true);
            if (!opts) {
                return 1;
            }
        }   break;
941 942 943
        case OPTION_IMAGE_OPTS:
            image_opts = true;
            break;
B
bellard 已提交
944 945
        }
    }
946 947 948 949 950 951

    /* Progress is not shown in Quiet mode */
    if (quiet) {
        progress = false;
    }

952
    if (optind != argc - 1) {
F
Fam Zheng 已提交
953
        error_exit("Expecting one image file name");
954
    }
B
bellard 已提交
955 956
    filename = argv[optind++];

957 958
    if (qemu_opts_foreach(&qemu_object_opts,
                          user_creatable_add_opts_foreach,
959
                          NULL, NULL)) {
960 961 962
        return 1;
    }

M
Max Reitz 已提交
963
    flags = BDRV_O_RDWR | BDRV_O_UNMAP;
964
    ret = bdrv_parse_cache_mode(cache, &flags, &writethrough);
965 966
    if (ret < 0) {
        error_report("Invalid cache option: %s", cache);
967
        return 1;
968 969
    }

970 971
    blk = img_open(image_opts, filename, fmt, flags, writethrough, quiet,
                   false);
972 973
    if (!blk) {
        return 1;
974
    }
975 976
    bs = blk_bs(blk);

977 978 979
    qemu_progress_init(progress, 1.f);
    qemu_progress_print(0.f, 100);

980 981 982
    if (base) {
        base_bs = bdrv_find_backing_image(bs, base);
        if (!base_bs) {
983 984 985
            error_setg(&local_err,
                       "Did not find '%s' in the backing chain of '%s'",
                       base, filename);
986 987 988 989 990 991
            goto done;
        }
    } else {
        /* This is different from QMP, which by default uses the deepest file in
         * the backing chain (i.e., the very base); however, the traditional
         * behavior of qemu-img commit is using the immediate backing file. */
992
        base_bs = backing_bs(bs);
993 994 995 996
        if (!base_bs) {
            error_setg(&local_err, "Image does not have a backing file");
            goto done;
        }
M
Max Reitz 已提交
997 998 999 1000 1001 1002 1003
    }

    cbi = (CommonBlockJobCBInfo){
        .errp = &local_err,
        .bs   = bs,
    };

1004 1005
    aio_context = bdrv_get_aio_context(bs);
    aio_context_acquire(aio_context);
1006
    commit_active_start("commit", bs, base_bs, BLOCK_JOB_DEFAULT, 0,
1007
                        BLOCKDEV_ON_ERROR_REPORT, NULL, common_block_job_cb,
1008
                        &cbi, false, &local_err);
1009
    aio_context_release(aio_context);
M
Max Reitz 已提交
1010 1011
    if (local_err) {
        goto done;
B
bellard 已提交
1012 1013
    }

1014 1015 1016 1017
    /* When the block job completes, the BlockBackend reference will point to
     * the old backing file. In order to avoid that the top image is already
     * deleted, so we can still empty it afterwards, increment the reference
     * counter here preemptively. */
M
Max Reitz 已提交
1018
    if (!drop) {
1019
        bdrv_ref(bs);
M
Max Reitz 已提交
1020 1021
    }

1022 1023
    job = block_job_get("commit");
    run_block_job(job, &local_err);
M
Max Reitz 已提交
1024 1025 1026 1027
    if (local_err) {
        goto unref_backing;
    }

1028 1029
    if (!drop && bs->drv->bdrv_make_empty) {
        ret = bs->drv->bdrv_make_empty(bs);
M
Max Reitz 已提交
1030 1031 1032 1033 1034 1035 1036 1037 1038
        if (ret) {
            error_setg_errno(&local_err, -ret, "Could not empty %s",
                             filename);
            goto unref_backing;
        }
    }

unref_backing:
    if (!drop) {
1039
        bdrv_unref(bs);
M
Max Reitz 已提交
1040
    }
M
Max Reitz 已提交
1041 1042

done:
1043 1044
    qemu_progress_end();

M
Markus Armbruster 已提交
1045
    blk_unref(blk);
M
Max Reitz 已提交
1046 1047

    if (local_err) {
1048
        error_report_err(local_err);
1049 1050
        return 1;
    }
M
Max Reitz 已提交
1051 1052

    qprintf(quiet, "Image committed.\n");
B
bellard 已提交
1053 1054 1055
    return 0;
}

1056 1057 1058 1059 1060 1061 1062
/*
 * Returns true iff the first sector pointed to by 'buf' contains at least
 * a non-NUL byte.
 *
 * 'pnum' is set to the number of sectors (including and immediately following
 * the first one) that are known to be in the same allocated/unallocated state.
 */
B
bellard 已提交
1063 1064
static int is_allocated_sectors(const uint8_t *buf, int n, int *pnum)
{
1065 1066
    bool is_zero;
    int i;
B
bellard 已提交
1067 1068 1069 1070 1071

    if (n <= 0) {
        *pnum = 0;
        return 0;
    }
1072
    is_zero = buffer_is_zero(buf, 512);
B
bellard 已提交
1073 1074
    for(i = 1; i < n; i++) {
        buf += 512;
1075
        if (is_zero != buffer_is_zero(buf, 512)) {
B
bellard 已提交
1076
            break;
1077
        }
B
bellard 已提交
1078 1079
    }
    *pnum = i;
1080
    return !is_zero;
B
bellard 已提交
1081 1082
}

1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124
/*
 * Like is_allocated_sectors, but if the buffer starts with a used sector,
 * up to 'min' consecutive sectors containing zeros are ignored. This avoids
 * breaking up write requests for only small sparse areas.
 */
static int is_allocated_sectors_min(const uint8_t *buf, int n, int *pnum,
    int min)
{
    int ret;
    int num_checked, num_used;

    if (n < min) {
        min = n;
    }

    ret = is_allocated_sectors(buf, n, pnum);
    if (!ret) {
        return ret;
    }

    num_used = *pnum;
    buf += BDRV_SECTOR_SIZE * *pnum;
    n -= *pnum;
    num_checked = num_used;

    while (n > 0) {
        ret = is_allocated_sectors(buf, n, pnum);

        buf += BDRV_SECTOR_SIZE * *pnum;
        n -= *pnum;
        num_checked += *pnum;
        if (ret) {
            num_used = num_checked;
        } else if (*pnum >= min) {
            break;
        }
    }

    *pnum = num_used;
    return 1;
}

K
Kevin Wolf 已提交
1125 1126 1127 1128 1129 1130 1131 1132 1133 1134
/*
 * Compares two buffers sector by sector. Returns 0 if the first sector of both
 * buffers matches, non-zero otherwise.
 *
 * pnum is set to the number of sectors (including and immediately following
 * the first one) that are known to have the same comparison result
 */
static int compare_sectors(const uint8_t *buf1, const uint8_t *buf2, int n,
    int *pnum)
{
1135 1136
    bool res;
    int i;
K
Kevin Wolf 已提交
1137 1138 1139 1140 1141 1142 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156

    if (n <= 0) {
        *pnum = 0;
        return 0;
    }

    res = !!memcmp(buf1, buf2, 512);
    for(i = 1; i < n; i++) {
        buf1 += 512;
        buf2 += 512;

        if (!!memcmp(buf1, buf2, 512) != res) {
            break;
        }
    }

    *pnum = i;
    return res;
}

1157
#define IO_BUF_SIZE (2 * 1024 * 1024)
B
bellard 已提交
1158

1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173 1174
static int64_t sectors_to_bytes(int64_t sectors)
{
    return sectors << BDRV_SECTOR_BITS;
}

static int64_t sectors_to_process(int64_t total, int64_t from)
{
    return MIN(total - from, IO_BUF_SIZE >> BDRV_SECTOR_BITS);
}

/*
 * Check if passed sectors are empty (not allocated or contain only 0 bytes)
 *
 * Returns 0 in case sectors are filled with 0, 1 if sectors contain non-zero
 * data and negative value on error.
 *
1175
 * @param blk:  BlockBackend for the image
1176 1177 1178 1179 1180 1181
 * @param sect_num: Number of first sector to check
 * @param sect_count: Number of sectors to check
 * @param filename: Name of disk file we are checking (logging purpose)
 * @param buffer: Allocated buffer for storing read data
 * @param quiet: Flag for quiet mode
 */
1182
static int check_empty_sectors(BlockBackend *blk, int64_t sect_num,
1183 1184 1185 1186
                               int sect_count, const char *filename,
                               uint8_t *buffer, bool quiet)
{
    int pnum, ret = 0;
1187 1188
    ret = blk_pread(blk, sect_num << BDRV_SECTOR_BITS, buffer,
                    sect_count << BDRV_SECTOR_BITS);
1189 1190 1191 1192 1193 1194 1195 1196 1197 1198 1199 1200 1201 1202 1203 1204 1205 1206 1207 1208 1209 1210 1211 1212
    if (ret < 0) {
        error_report("Error while reading offset %" PRId64 " of %s: %s",
                     sectors_to_bytes(sect_num), filename, strerror(-ret));
        return ret;
    }
    ret = is_allocated_sectors(buffer, sect_count, &pnum);
    if (ret || pnum != sect_count) {
        qprintf(quiet, "Content mismatch at offset %" PRId64 "!\n",
                sectors_to_bytes(ret ? sect_num : sect_num + pnum));
        return 1;
    }

    return 0;
}

/*
 * Compares two images. Exit codes:
 *
 * 0 - Images are identical
 * 1 - Images differ
 * >1 - Error occurred
 */
static int img_compare(int argc, char **argv)
{
1213
    const char *fmt1 = NULL, *fmt2 = NULL, *cache, *filename1, *filename2;
M
Markus Armbruster 已提交
1214
    BlockBackend *blk1, *blk2;
1215 1216 1217 1218 1219 1220 1221
    BlockDriverState *bs1, *bs2;
    int64_t total_sectors1, total_sectors2;
    uint8_t *buf1 = NULL, *buf2 = NULL;
    int pnum1, pnum2;
    int allocated1, allocated2;
    int ret = 0; /* return value - 0 Ident, 1 Different, >1 Error */
    bool progress = false, quiet = false, strict = false;
1222
    int flags;
1223
    bool writethrough;
1224 1225 1226 1227 1228
    int64_t total_sectors;
    int64_t sector_num = 0;
    int64_t nb_sectors;
    int c, pnum;
    uint64_t progress_base;
1229
    bool image_opts = false;
1230
    bool force_share = false;
1231

1232
    cache = BDRV_DEFAULT_CACHE;
1233
    for (;;) {
1234 1235 1236
        static const struct option long_options[] = {
            {"help", no_argument, 0, 'h'},
            {"object", required_argument, 0, OPTION_OBJECT},
1237
            {"image-opts", no_argument, 0, OPTION_IMAGE_OPTS},
1238
            {"force-share", no_argument, 0, 'U'},
1239 1240
            {0, 0, 0, 0}
        };
1241
        c = getopt_long(argc, argv, ":hf:F:T:pqsU",
1242
                        long_options, NULL);
1243 1244 1245 1246
        if (c == -1) {
            break;
        }
        switch (c) {
1247 1248 1249
        case ':':
            missing_argument(argv[optind - 1]);
            break;
1250
        case '?':
1251 1252
            unrecognized_option(argv[optind - 1]);
            break;
1253 1254 1255 1256 1257 1258 1259 1260 1261
        case 'h':
            help();
            break;
        case 'f':
            fmt1 = optarg;
            break;
        case 'F':
            fmt2 = optarg;
            break;
1262 1263 1264
        case 'T':
            cache = optarg;
            break;
1265 1266 1267 1268 1269 1270 1271 1272 1273
        case 'p':
            progress = true;
            break;
        case 'q':
            quiet = true;
            break;
        case 's':
            strict = true;
            break;
1274 1275 1276
        case 'U':
            force_share = true;
            break;
1277 1278 1279 1280 1281 1282 1283 1284 1285
        case OPTION_OBJECT: {
            QemuOpts *opts;
            opts = qemu_opts_parse_noisily(&qemu_object_opts,
                                           optarg, true);
            if (!opts) {
                ret = 2;
                goto out4;
            }
        }   break;
1286 1287 1288
        case OPTION_IMAGE_OPTS:
            image_opts = true;
            break;
1289 1290 1291 1292 1293 1294 1295 1296 1297
        }
    }

    /* Progress is not shown in Quiet mode */
    if (quiet) {
        progress = false;
    }


1298
    if (optind != argc - 2) {
F
Fam Zheng 已提交
1299
        error_exit("Expecting two image file names");
1300 1301 1302 1303
    }
    filename1 = argv[optind++];
    filename2 = argv[optind++];

1304 1305
    if (qemu_opts_foreach(&qemu_object_opts,
                          user_creatable_add_opts_foreach,
1306
                          NULL, NULL)) {
1307 1308 1309 1310
        ret = 2;
        goto out4;
    }

1311 1312 1313
    /* Initialize before goto out */
    qemu_progress_init(progress, 2.0);

1314 1315
    flags = 0;
    ret = bdrv_parse_cache_mode(cache, &flags, &writethrough);
1316 1317 1318 1319 1320 1321
    if (ret < 0) {
        error_report("Invalid source cache option: %s", cache);
        ret = 2;
        goto out3;
    }

1322 1323
    blk1 = img_open(image_opts, filename1, fmt1, flags, writethrough, quiet,
                    force_share);
1324
    if (!blk1) {
1325
        ret = 2;
1326
        goto out3;
1327 1328
    }

1329 1330
    blk2 = img_open(image_opts, filename2, fmt2, flags, writethrough, quiet,
                    force_share);
1331
    if (!blk2) {
1332
        ret = 2;
1333
        goto out2;
1334
    }
1335
    bs1 = blk_bs(blk1);
1336
    bs2 = blk_bs(blk2);
1337

1338 1339 1340
    buf1 = blk_blockalign(blk1, IO_BUF_SIZE);
    buf2 = blk_blockalign(blk2, IO_BUF_SIZE);
    total_sectors1 = blk_nb_sectors(blk1);
1341 1342 1343 1344 1345 1346
    if (total_sectors1 < 0) {
        error_report("Can't get size of %s: %s",
                     filename1, strerror(-total_sectors1));
        ret = 4;
        goto out;
    }
1347
    total_sectors2 = blk_nb_sectors(blk2);
1348 1349 1350 1351 1352 1353
    if (total_sectors2 < 0) {
        error_report("Can't get size of %s: %s",
                     filename2, strerror(-total_sectors2));
        ret = 4;
        goto out;
    }
1354 1355 1356 1357 1358 1359 1360 1361 1362 1363 1364 1365
    total_sectors = MIN(total_sectors1, total_sectors2);
    progress_base = MAX(total_sectors1, total_sectors2);

    qemu_progress_print(0, 100);

    if (strict && total_sectors1 != total_sectors2) {
        ret = 1;
        qprintf(quiet, "Strict mode: Image size mismatch!\n");
        goto out;
    }

    for (;;) {
1366
        int64_t status1, status2;
1367 1368
        BlockDriverState *file;

1369 1370 1371 1372
        nb_sectors = sectors_to_process(total_sectors, sector_num);
        if (nb_sectors <= 0) {
            break;
        }
1373 1374
        status1 = bdrv_get_block_status_above(bs1, NULL, sector_num,
                                              total_sectors1 - sector_num,
1375
                                              &pnum1, &file);
1376
        if (status1 < 0) {
1377 1378 1379 1380
            ret = 3;
            error_report("Sector allocation test failed for %s", filename1);
            goto out;
        }
1381
        allocated1 = status1 & BDRV_BLOCK_ALLOCATED;
1382

1383 1384
        status2 = bdrv_get_block_status_above(bs2, NULL, sector_num,
                                              total_sectors2 - sector_num,
1385
                                              &pnum2, &file);
1386
        if (status2 < 0) {
1387 1388 1389 1390
            ret = 3;
            error_report("Sector allocation test failed for %s", filename2);
            goto out;
        }
1391 1392 1393 1394 1395 1396 1397
        allocated2 = status2 & BDRV_BLOCK_ALLOCATED;
        if (pnum1) {
            nb_sectors = MIN(nb_sectors, pnum1);
        }
        if (pnum2) {
            nb_sectors = MIN(nb_sectors, pnum2);
        }
1398

1399 1400 1401 1402 1403 1404 1405 1406 1407 1408 1409 1410 1411
        if (strict) {
            if ((status1 & ~BDRV_BLOCK_OFFSET_MASK) !=
                (status2 & ~BDRV_BLOCK_OFFSET_MASK)) {
                ret = 1;
                qprintf(quiet, "Strict mode: Offset %" PRId64
                        " block status mismatch!\n",
                        sectors_to_bytes(sector_num));
                goto out;
            }
        }
        if ((status1 & BDRV_BLOCK_ZERO) && (status2 & BDRV_BLOCK_ZERO)) {
            nb_sectors = MIN(pnum1, pnum2);
        } else if (allocated1 == allocated2) {
1412
            if (allocated1) {
1413 1414
                ret = blk_pread(blk1, sector_num << BDRV_SECTOR_BITS, buf1,
                                nb_sectors << BDRV_SECTOR_BITS);
1415 1416 1417 1418 1419 1420 1421
                if (ret < 0) {
                    error_report("Error while reading offset %" PRId64 " of %s:"
                                 " %s", sectors_to_bytes(sector_num), filename1,
                                 strerror(-ret));
                    ret = 4;
                    goto out;
                }
1422 1423
                ret = blk_pread(blk2, sector_num << BDRV_SECTOR_BITS, buf2,
                                nb_sectors << BDRV_SECTOR_BITS);
1424 1425 1426 1427 1428 1429 1430 1431 1432 1433 1434 1435
                if (ret < 0) {
                    error_report("Error while reading offset %" PRId64
                                 " of %s: %s", sectors_to_bytes(sector_num),
                                 filename2, strerror(-ret));
                    ret = 4;
                    goto out;
                }
                ret = compare_sectors(buf1, buf2, nb_sectors, &pnum);
                if (ret || pnum != nb_sectors) {
                    qprintf(quiet, "Content mismatch at offset %" PRId64 "!\n",
                            sectors_to_bytes(
                                ret ? sector_num : sector_num + pnum));
1436
                    ret = 1;
1437 1438 1439 1440 1441 1442
                    goto out;
                }
            }
        } else {

            if (allocated1) {
1443
                ret = check_empty_sectors(blk1, sector_num, nb_sectors,
1444 1445
                                          filename1, buf1, quiet);
            } else {
1446
                ret = check_empty_sectors(blk2, sector_num, nb_sectors,
1447 1448 1449 1450 1451 1452
                                          filename2, buf1, quiet);
            }
            if (ret) {
                if (ret < 0) {
                    error_report("Error while reading offset %" PRId64 ": %s",
                                 sectors_to_bytes(sector_num), strerror(-ret));
1453
                    ret = 4;
1454 1455 1456 1457 1458 1459 1460 1461 1462
                }
                goto out;
            }
        }
        sector_num += nb_sectors;
        qemu_progress_print(((float) nb_sectors / progress_base)*100, 100);
    }

    if (total_sectors1 != total_sectors2) {
1463
        BlockBackend *blk_over;
1464 1465 1466 1467 1468 1469
        int64_t total_sectors_over;
        const char *filename_over;

        qprintf(quiet, "Warning: Image size mismatch!\n");
        if (total_sectors1 > total_sectors2) {
            total_sectors_over = total_sectors1;
1470
            blk_over = blk1;
1471 1472 1473
            filename_over = filename1;
        } else {
            total_sectors_over = total_sectors2;
1474
            blk_over = blk2;
1475 1476 1477 1478 1479 1480 1481 1482
            filename_over = filename2;
        }

        for (;;) {
            nb_sectors = sectors_to_process(total_sectors_over, sector_num);
            if (nb_sectors <= 0) {
                break;
            }
1483
            ret = bdrv_is_allocated_above(blk_bs(blk_over), NULL, sector_num,
1484 1485 1486 1487 1488 1489 1490 1491 1492 1493
                                          nb_sectors, &pnum);
            if (ret < 0) {
                ret = 3;
                error_report("Sector allocation test failed for %s",
                             filename_over);
                goto out;

            }
            nb_sectors = pnum;
            if (ret) {
1494
                ret = check_empty_sectors(blk_over, sector_num, nb_sectors,
1495 1496 1497 1498 1499 1500
                                          filename_over, buf1, quiet);
                if (ret) {
                    if (ret < 0) {
                        error_report("Error while reading offset %" PRId64
                                     " of %s: %s", sectors_to_bytes(sector_num),
                                     filename_over, strerror(-ret));
1501
                        ret = 4;
1502 1503 1504 1505 1506 1507 1508 1509 1510 1511 1512 1513 1514 1515 1516
                    }
                    goto out;
                }
            }
            sector_num += nb_sectors;
            qemu_progress_print(((float) nb_sectors / progress_base)*100, 100);
        }
    }

    qprintf(quiet, "Images are identical.\n");
    ret = 0;

out:
    qemu_vfree(buf1);
    qemu_vfree(buf2);
M
Markus Armbruster 已提交
1517
    blk_unref(blk2);
1518
out2:
M
Markus Armbruster 已提交
1519
    blk_unref(blk1);
1520 1521
out3:
    qemu_progress_end();
1522
out4:
1523 1524 1525
    return ret;
}

1526 1527 1528 1529 1530 1531
enum ImgConvertBlockStatus {
    BLK_DATA,
    BLK_ZERO,
    BLK_BACKING_FILE,
};

P
Peter Lieven 已提交
1532 1533
#define MAX_COROUTINES 16

1534 1535 1536
typedef struct ImgConvertState {
    BlockBackend **src;
    int64_t *src_sectors;
P
Peter Lieven 已提交
1537
    int src_num;
1538 1539
    int64_t total_sectors;
    int64_t allocated_sectors;
P
Peter Lieven 已提交
1540 1541 1542
    int64_t allocated_done;
    int64_t sector_num;
    int64_t wr_offs;
1543 1544 1545 1546 1547 1548
    enum ImgConvertBlockStatus status;
    int64_t sector_next_status;
    BlockBackend *target;
    bool has_zero_init;
    bool compressed;
    bool target_has_backing;
P
Peter Lieven 已提交
1549
    bool wr_in_order;
1550 1551 1552
    int min_sparse;
    size_t cluster_sectors;
    size_t buf_sectors;
P
Peter Lieven 已提交
1553
    long num_coroutines;
P
Peter Lieven 已提交
1554 1555 1556 1557 1558
    int running_coroutines;
    Coroutine *co[MAX_COROUTINES];
    int64_t wait_sector_num[MAX_COROUTINES];
    CoMutex lock;
    int ret;
1559 1560
} ImgConvertState;

P
Peter Lieven 已提交
1561 1562
static void convert_select_part(ImgConvertState *s, int64_t sector_num,
                                int *src_cur, int64_t *src_cur_offset)
1563
{
P
Peter Lieven 已提交
1564 1565 1566 1567 1568 1569
    *src_cur = 0;
    *src_cur_offset = 0;
    while (sector_num - *src_cur_offset >= s->src_sectors[*src_cur]) {
        *src_cur_offset += s->src_sectors[*src_cur];
        (*src_cur)++;
        assert(*src_cur < s->src_num);
1570 1571 1572 1573 1574
    }
}

static int convert_iteration_sectors(ImgConvertState *s, int64_t sector_num)
{
P
Peter Lieven 已提交
1575 1576
    int64_t ret, src_cur_offset;
    int n, src_cur;
1577

P
Peter Lieven 已提交
1578
    convert_select_part(s, sector_num, &src_cur, &src_cur_offset);
1579 1580 1581 1582 1583

    assert(s->total_sectors > sector_num);
    n = MIN(s->total_sectors - sector_num, BDRV_REQUEST_MAX_SECTORS);

    if (s->sector_next_status <= sector_num) {
1584
        BlockDriverState *file;
1585 1586 1587 1588 1589 1590 1591 1592 1593
        if (s->target_has_backing) {
            ret = bdrv_get_block_status(blk_bs(s->src[src_cur]),
                                        sector_num - src_cur_offset,
                                        n, &n, &file);
        } else {
            ret = bdrv_get_block_status_above(blk_bs(s->src[src_cur]), NULL,
                                              sector_num - src_cur_offset,
                                              n, &n, &file);
        }
1594 1595 1596 1597 1598 1599 1600 1601 1602
        if (ret < 0) {
            return ret;
        }

        if (ret & BDRV_BLOCK_ZERO) {
            s->status = BLK_ZERO;
        } else if (ret & BDRV_BLOCK_DATA) {
            s->status = BLK_DATA;
        } else {
1603
            s->status = s->target_has_backing ? BLK_BACKING_FILE : BLK_DATA;
1604 1605 1606 1607 1608 1609 1610 1611 1612 1613 1614 1615 1616 1617 1618 1619 1620 1621 1622 1623 1624 1625 1626 1627 1628
        }

        s->sector_next_status = sector_num + n;
    }

    n = MIN(n, s->sector_next_status - sector_num);
    if (s->status == BLK_DATA) {
        n = MIN(n, s->buf_sectors);
    }

    /* We need to write complete clusters for compressed images, so if an
     * unallocated area is shorter than that, we must consider the whole
     * cluster allocated. */
    if (s->compressed) {
        if (n < s->cluster_sectors) {
            n = MIN(s->cluster_sectors, s->total_sectors - sector_num);
            s->status = BLK_DATA;
        } else {
            n = QEMU_ALIGN_DOWN(n, s->cluster_sectors);
        }
    }

    return n;
}

P
Peter Lieven 已提交
1629 1630
static int coroutine_fn convert_co_read(ImgConvertState *s, int64_t sector_num,
                                        int nb_sectors, uint8_t *buf)
1631
{
P
Peter Lieven 已提交
1632 1633 1634
    int n, ret;
    QEMUIOVector qiov;
    struct iovec iov;
1635 1636 1637 1638

    assert(nb_sectors <= s->buf_sectors);
    while (nb_sectors > 0) {
        BlockBackend *blk;
P
Peter Lieven 已提交
1639 1640
        int src_cur;
        int64_t bs_sectors, src_cur_offset;
1641 1642 1643 1644

        /* In the case of compression with multiple source files, we can get a
         * nb_sectors that spreads into the next part. So we must be able to
         * read across multiple BDSes for one convert_read() call. */
P
Peter Lieven 已提交
1645 1646 1647 1648 1649 1650 1651 1652 1653 1654 1655 1656
        convert_select_part(s, sector_num, &src_cur, &src_cur_offset);
        blk = s->src[src_cur];
        bs_sectors = s->src_sectors[src_cur];

        n = MIN(nb_sectors, bs_sectors - (sector_num - src_cur_offset));
        iov.iov_base = buf;
        iov.iov_len = n << BDRV_SECTOR_BITS;
        qemu_iovec_init_external(&qiov, &iov, 1);

        ret = blk_co_preadv(
                blk, (sector_num - src_cur_offset) << BDRV_SECTOR_BITS,
                n << BDRV_SECTOR_BITS, &qiov, 0);
1657 1658 1659 1660 1661 1662 1663 1664 1665 1666 1667 1668
        if (ret < 0) {
            return ret;
        }

        sector_num += n;
        nb_sectors -= n;
        buf += n * BDRV_SECTOR_SIZE;
    }

    return 0;
}

P
Peter Lieven 已提交
1669 1670 1671 1672

static int coroutine_fn convert_co_write(ImgConvertState *s, int64_t sector_num,
                                         int nb_sectors, uint8_t *buf,
                                         enum ImgConvertBlockStatus status)
1673 1674
{
    int ret;
P
Peter Lieven 已提交
1675 1676
    QEMUIOVector qiov;
    struct iovec iov;
1677 1678 1679

    while (nb_sectors > 0) {
        int n = nb_sectors;
1680 1681
        BdrvRequestFlags flags = s->compressed ? BDRV_REQ_WRITE_COMPRESSED : 0;

P
Peter Lieven 已提交
1682
        switch (status) {
1683 1684 1685 1686 1687 1688 1689 1690
        case BLK_BACKING_FILE:
            /* If we have a backing file, leave clusters unallocated that are
             * unallocated in the source image, so that the backing file is
             * visible at the respective offset. */
            assert(s->target_has_backing);
            break;

        case BLK_DATA:
1691 1692 1693 1694 1695 1696
            /* If we're told to keep the target fully allocated (-S 0) or there
             * is real non-zero data, we must write it. Otherwise we can treat
             * it as zero sectors.
             * Compressed clusters need to be written as a whole, so in that
             * case we can only save the write if the buffer is completely
             * zeroed. */
1697
            if (!s->min_sparse ||
1698 1699 1700 1701
                (!s->compressed &&
                 is_allocated_sectors_min(buf, n, &n, s->min_sparse)) ||
                (s->compressed &&
                 !buffer_is_zero(buf, n * BDRV_SECTOR_SIZE)))
1702
            {
P
Peter Lieven 已提交
1703 1704 1705 1706 1707
                iov.iov_base = buf;
                iov.iov_len = n << BDRV_SECTOR_BITS;
                qemu_iovec_init_external(&qiov, &iov, 1);

                ret = blk_co_pwritev(s->target, sector_num << BDRV_SECTOR_BITS,
1708
                                     n << BDRV_SECTOR_BITS, &qiov, flags);
1709 1710 1711 1712 1713 1714 1715 1716 1717
                if (ret < 0) {
                    return ret;
                }
                break;
            }
            /* fall-through */

        case BLK_ZERO:
            if (s->has_zero_init) {
1718
                assert(!s->target_has_backing);
1719 1720
                break;
            }
P
Peter Lieven 已提交
1721 1722 1723
            ret = blk_co_pwrite_zeroes(s->target,
                                       sector_num << BDRV_SECTOR_BITS,
                                       n << BDRV_SECTOR_BITS, 0);
1724 1725 1726 1727 1728 1729 1730 1731 1732 1733 1734 1735 1736 1737
            if (ret < 0) {
                return ret;
            }
            break;
        }

        sector_num += n;
        nb_sectors -= n;
        buf += n * BDRV_SECTOR_SIZE;
    }

    return 0;
}

P
Peter Lieven 已提交
1738
static void coroutine_fn convert_co_do_copy(void *opaque)
1739
{
P
Peter Lieven 已提交
1740
    ImgConvertState *s = opaque;
1741
    uint8_t *buf = NULL;
P
Peter Lieven 已提交
1742 1743 1744 1745 1746 1747 1748 1749 1750 1751 1752 1753 1754 1755 1756 1757 1758 1759 1760 1761 1762 1763
    int ret, i;
    int index = -1;

    for (i = 0; i < s->num_coroutines; i++) {
        if (s->co[i] == qemu_coroutine_self()) {
            index = i;
            break;
        }
    }
    assert(index >= 0);

    s->running_coroutines++;
    buf = blk_blockalign(s->target, s->buf_sectors * BDRV_SECTOR_SIZE);

    while (1) {
        int n;
        int64_t sector_num;
        enum ImgConvertBlockStatus status;

        qemu_co_mutex_lock(&s->lock);
        if (s->ret != -EINPROGRESS || s->sector_num >= s->total_sectors) {
            qemu_co_mutex_unlock(&s->lock);
1764
            break;
P
Peter Lieven 已提交
1765 1766 1767 1768 1769
        }
        n = convert_iteration_sectors(s, s->sector_num);
        if (n < 0) {
            qemu_co_mutex_unlock(&s->lock);
            s->ret = n;
1770
            break;
P
Peter Lieven 已提交
1771 1772 1773 1774 1775 1776 1777 1778 1779 1780 1781 1782 1783 1784 1785 1786 1787 1788 1789 1790 1791 1792 1793 1794 1795 1796 1797 1798 1799 1800 1801 1802
        }
        /* save current sector and allocation status to local variables */
        sector_num = s->sector_num;
        status = s->status;
        if (!s->min_sparse && s->status == BLK_ZERO) {
            n = MIN(n, s->buf_sectors);
        }
        /* increment global sector counter so that other coroutines can
         * already continue reading beyond this request */
        s->sector_num += n;
        qemu_co_mutex_unlock(&s->lock);

        if (status == BLK_DATA || (!s->min_sparse && status == BLK_ZERO)) {
            s->allocated_done += n;
            qemu_progress_print(100.0 * s->allocated_done /
                                        s->allocated_sectors, 0);
        }

        if (status == BLK_DATA) {
            ret = convert_co_read(s, sector_num, n, buf);
            if (ret < 0) {
                error_report("error while reading sector %" PRId64
                             ": %s", sector_num, strerror(-ret));
                s->ret = ret;
            }
        } else if (!s->min_sparse && status == BLK_ZERO) {
            status = BLK_DATA;
            memset(buf, 0x00, n * BDRV_SECTOR_SIZE);
        }

        if (s->wr_in_order) {
            /* keep writes in order */
1803
            while (s->wr_offs != sector_num && s->ret == -EINPROGRESS) {
P
Peter Lieven 已提交
1804 1805 1806 1807 1808 1809
                s->wait_sector_num[index] = sector_num;
                qemu_coroutine_yield();
            }
            s->wait_sector_num[index] = -1;
        }

1810 1811 1812 1813 1814 1815 1816
        if (s->ret == -EINPROGRESS) {
            ret = convert_co_write(s, sector_num, n, buf, status);
            if (ret < 0) {
                error_report("error while writing sector %" PRId64
                             ": %s", sector_num, strerror(-ret));
                s->ret = ret;
            }
P
Peter Lieven 已提交
1817 1818 1819 1820 1821 1822 1823 1824 1825 1826 1827 1828 1829 1830 1831 1832 1833 1834 1835 1836 1837 1838 1839 1840 1841 1842 1843 1844 1845 1846 1847 1848 1849
        }

        if (s->wr_in_order) {
            /* reenter the coroutine that might have waited
             * for this write to complete */
            s->wr_offs = sector_num + n;
            for (i = 0; i < s->num_coroutines; i++) {
                if (s->co[i] && s->wait_sector_num[i] == s->wr_offs) {
                    /*
                     * A -> B -> A cannot occur because A has
                     * s->wait_sector_num[i] == -1 during A -> B.  Therefore
                     * B will never enter A during this time window.
                     */
                    qemu_coroutine_enter(s->co[i]);
                    break;
                }
            }
        }
    }

    qemu_vfree(buf);
    s->co[index] = NULL;
    s->running_coroutines--;
    if (!s->running_coroutines && s->ret == -EINPROGRESS) {
        /* the convert job finished successfully */
        s->ret = 0;
    }
}

static int convert_do_copy(ImgConvertState *s)
{
    int ret, i, n;
    int64_t sector_num = 0;
1850 1851 1852 1853 1854 1855 1856 1857 1858

    /* Check whether we have zero initialisation or can get it efficiently */
    s->has_zero_init = s->min_sparse && !s->target_has_backing
                     ? bdrv_has_zero_init(blk_bs(s->target))
                     : false;

    if (!s->has_zero_init && !s->target_has_backing &&
        bdrv_can_write_zeroes_with_unmap(blk_bs(s->target)))
    {
1859
        ret = blk_make_zero(s->target, BDRV_REQ_MAY_UNMAP);
1860 1861 1862 1863 1864 1865 1866 1867 1868 1869
        if (ret == 0) {
            s->has_zero_init = true;
        }
    }

    /* Allocate buffer for copied data. For compressed images, only one cluster
     * can be copied at a time. */
    if (s->compressed) {
        if (s->cluster_sectors <= 0 || s->cluster_sectors > s->buf_sectors) {
            error_report("invalid cluster size");
P
Peter Lieven 已提交
1870
            return -EINVAL;
1871 1872 1873 1874 1875 1876 1877
        }
        s->buf_sectors = s->cluster_sectors;
    }

    while (sector_num < s->total_sectors) {
        n = convert_iteration_sectors(s, sector_num);
        if (n < 0) {
P
Peter Lieven 已提交
1878
            return n;
1879
        }
1880 1881
        if (s->status == BLK_DATA || (!s->min_sparse && s->status == BLK_ZERO))
        {
1882 1883 1884 1885 1886 1887 1888
            s->allocated_sectors += n;
        }
        sector_num += n;
    }

    /* Do the copy */
    s->sector_next_status = 0;
P
Peter Lieven 已提交
1889
    s->ret = -EINPROGRESS;
1890

P
Peter Lieven 已提交
1891 1892 1893 1894 1895 1896
    qemu_co_mutex_init(&s->lock);
    for (i = 0; i < s->num_coroutines; i++) {
        s->co[i] = qemu_coroutine_create(convert_co_do_copy, s);
        s->wait_sector_num[i] = -1;
        qemu_coroutine_enter(s->co[i]);
    }
1897

1898
    while (s->running_coroutines) {
P
Peter Lieven 已提交
1899
        main_loop_wait(false);
1900 1901
    }

P
Peter Lieven 已提交
1902
    if (s->compressed && !s->ret) {
1903
        /* signal EOF to align */
1904
        ret = blk_pwrite_compressed(s->target, 0, NULL, 0);
1905
        if (ret < 0) {
P
Peter Lieven 已提交
1906
            return ret;
1907 1908 1909
        }
    }

P
Peter Lieven 已提交
1910
    return s->ret;
1911 1912
}

B
bellard 已提交
1913 1914
static int img_convert(int argc, char **argv)
{
P
Peter Lieven 已提交
1915 1916 1917 1918
    int c, bs_i, flags, src_flags = 0;
    const char *fmt = NULL, *out_fmt = "raw", *cache = "unsafe",
               *src_cache = BDRV_DEFAULT_CACHE, *out_baseimg = NULL,
               *out_filename, *out_baseimg_param, *snapshot_name = NULL;
1919
    BlockDriver *drv, *proto_drv;
B
bellard 已提交
1920
    BlockDriverInfo bdi;
P
Peter Lieven 已提交
1921 1922
    BlockDriverState *out_bs;
    QemuOpts *opts = NULL, *sn_opts = NULL;
1923
    QemuOptsList *create_opts = NULL;
1924
    char *options = NULL;
1925
    Error *local_err = NULL;
P
Peter Lieven 已提交
1926 1927 1928
    bool writethrough, src_writethrough, quiet = false, image_opts = false,
         skip_create = false, progress = false;
    int64_t ret = -EINVAL;
1929
    bool force_share = false;
P
Peter Lieven 已提交
1930 1931 1932 1933 1934 1935 1936 1937

    ImgConvertState s = (ImgConvertState) {
        /* Need at least 4k of zeros for sparse detection */
        .min_sparse         = 8,
        .buf_sectors        = IO_BUF_SIZE / BDRV_SECTOR_SIZE,
        .wr_in_order        = true,
        .num_coroutines     = 8,
    };
B
bellard 已提交
1938 1939

    for(;;) {
1940 1941 1942
        static const struct option long_options[] = {
            {"help", no_argument, 0, 'h'},
            {"object", required_argument, 0, OPTION_OBJECT},
1943
            {"image-opts", no_argument, 0, OPTION_IMAGE_OPTS},
1944
            {"force-share", no_argument, 0, 'U'},
1945 1946
            {0, 0, 0, 0}
        };
1947
        c = getopt_long(argc, argv, ":hf:O:B:ce6o:s:l:S:pt:T:qnm:WU",
1948
                        long_options, NULL);
1949
        if (c == -1) {
B
bellard 已提交
1950
            break;
1951
        }
B
bellard 已提交
1952
        switch(c) {
1953 1954 1955
        case ':':
            missing_argument(argv[optind - 1]);
            break;
J
Jes Sorensen 已提交
1956
        case '?':
1957 1958
            unrecognized_option(argv[optind - 1]);
            break;
B
bellard 已提交
1959 1960 1961 1962 1963 1964 1965 1966 1967
        case 'h':
            help();
            break;
        case 'f':
            fmt = optarg;
            break;
        case 'O':
            out_fmt = optarg;
            break;
1968 1969 1970
        case 'B':
            out_baseimg = optarg;
            break;
B
bellard 已提交
1971
        case 'c':
P
Peter Lieven 已提交
1972
            s.compressed = true;
B
bellard 已提交
1973 1974
            break;
        case 'e':
1975
            error_report("option -e is deprecated, please use \'-o "
1976
                  "encryption\' instead!");
1977
            goto fail_getopt;
1978
        case '6':
1979
            error_report("option -6 is deprecated, please use \'-o "
1980
                  "compat6\' instead!");
1981
            goto fail_getopt;
1982
        case 'o':
1983 1984
            if (!is_valid_option_list(optarg)) {
                error_report("Invalid option list: %s", optarg);
1985
                goto fail_getopt;
1986 1987 1988 1989 1990 1991 1992 1993
            }
            if (!options) {
                options = g_strdup(optarg);
            } else {
                char *old_options = options;
                options = g_strdup_printf("%s,%s", options, optarg);
                g_free(old_options);
            }
1994
            break;
E
edison 已提交
1995 1996 1997
        case 's':
            snapshot_name = optarg;
            break;
1998 1999
        case 'l':
            if (strstart(optarg, SNAPSHOT_OPT_BASE, NULL)) {
2000 2001
                sn_opts = qemu_opts_parse_noisily(&internal_snapshot_opts,
                                                  optarg, false);
2002 2003 2004
                if (!sn_opts) {
                    error_report("Failed in parsing snapshot param '%s'",
                                 optarg);
2005
                    goto fail_getopt;
2006 2007 2008 2009 2010
                }
            } else {
                snapshot_name = optarg;
            }
            break;
2011 2012 2013
        case 'S':
        {
            int64_t sval;
2014 2015 2016

            sval = cvtnum(optarg);
            if (sval < 0) {
2017
                error_report("Invalid minimum zero buffer size for sparse output specified");
2018
                goto fail_getopt;
2019 2020
            }

P
Peter Lieven 已提交
2021
            s.min_sparse = sval / BDRV_SECTOR_SIZE;
2022 2023
            break;
        }
2024
        case 'p':
P
Peter Lieven 已提交
2025
            progress = true;
2026
            break;
2027 2028 2029
        case 't':
            cache = optarg;
            break;
2030 2031 2032
        case 'T':
            src_cache = optarg;
            break;
2033 2034 2035
        case 'q':
            quiet = true;
            break;
2036
        case 'n':
P
Peter Lieven 已提交
2037
            skip_create = true;
2038
            break;
P
Peter Lieven 已提交
2039
        case 'm':
P
Peter Lieven 已提交
2040 2041
            if (qemu_strtol(optarg, NULL, 0, &s.num_coroutines) ||
                s.num_coroutines < 1 || s.num_coroutines > MAX_COROUTINES) {
P
Peter Lieven 已提交
2042 2043 2044 2045 2046 2047
                error_report("Invalid number of coroutines. Allowed number of"
                             " coroutines is between 1 and %d", MAX_COROUTINES);
                goto fail_getopt;
            }
            break;
        case 'W':
P
Peter Lieven 已提交
2048
            s.wr_in_order = false;
P
Peter Lieven 已提交
2049
            break;
2050 2051 2052
        case 'U':
            force_share = true;
            break;
2053 2054 2055 2056 2057
        case OPTION_OBJECT: {
            QemuOpts *object_opts;
            object_opts = qemu_opts_parse_noisily(&qemu_object_opts,
                                                  optarg, true);
            if (!object_opts) {
2058 2059 2060
                goto fail_getopt;
            }
            break;
2061
        }
2062 2063 2064
        case OPTION_IMAGE_OPTS:
            image_opts = true;
            break;
B
bellard 已提交
2065 2066
        }
    }
2067

2068 2069
    if (qemu_opts_foreach(&qemu_object_opts,
                          user_creatable_add_opts_foreach,
2070
                          NULL, NULL)) {
2071 2072 2073
        goto fail_getopt;
    }

P
Peter Lieven 已提交
2074
    if (!s.wr_in_order && s.compressed) {
P
Peter Lieven 已提交
2075 2076 2077 2078
        error_report("Out of order write and compress are mutually exclusive");
        goto fail_getopt;
    }

P
Peter Lieven 已提交
2079 2080
    s.src_num = argc - optind - 1;
    out_filename = s.src_num >= 1 ? argv[argc - 1] : NULL;
2081

2082
    if (options && has_help_option(options)) {
2083
        ret = print_block_option_help(out_filename, out_fmt);
P
Peter Lieven 已提交
2084
        goto fail_getopt;
2085 2086
    }

P
Peter Lieven 已提交
2087 2088 2089
    if (s.src_num < 1) {
        error_report("Must specify image file name");
        goto fail_getopt;
2090 2091 2092
    }


P
Peter Lieven 已提交
2093
    /* ret is still -EINVAL until here */
2094
    ret = bdrv_parse_cache_mode(src_cache, &src_flags, &src_writethrough);
2095 2096
    if (ret < 0) {
        error_report("Invalid source cache option: %s", src_cache);
P
Peter Lieven 已提交
2097
        goto fail_getopt;
2098 2099
    }

P
Peter Lieven 已提交
2100 2101 2102 2103 2104
    /* Initialize before goto out */
    if (quiet) {
        progress = false;
    }
    qemu_progress_init(progress, 1.0);
2105 2106
    qemu_progress_print(0, 100);

P
Peter Lieven 已提交
2107 2108
    s.src = g_new0(BlockBackend *, s.src_num);
    s.src_sectors = g_new(int64_t, s.src_num);
2109

P
Peter Lieven 已提交
2110 2111
    for (bs_i = 0; bs_i < s.src_num; bs_i++) {
        s.src[bs_i] = img_open(image_opts, argv[optind + bs_i],
2112 2113
                               fmt, src_flags, src_writethrough, quiet,
                               force_share);
P
Peter Lieven 已提交
2114
        if (!s.src[bs_i]) {
2115 2116 2117
            ret = -1;
            goto out;
        }
P
Peter Lieven 已提交
2118 2119
        s.src_sectors[bs_i] = blk_nb_sectors(s.src[bs_i]);
        if (s.src_sectors[bs_i] < 0) {
2120
            error_report("Could not get size of %s: %s",
P
Peter Lieven 已提交
2121
                         argv[optind + bs_i], strerror(-s.src_sectors[bs_i]));
2122 2123 2124
            ret = -1;
            goto out;
        }
P
Peter Lieven 已提交
2125
        s.total_sectors += s.src_sectors[bs_i];
2126
    }
B
bellard 已提交
2127

2128
    if (sn_opts) {
P
Peter Lieven 已提交
2129
        bdrv_snapshot_load_tmp(blk_bs(s.src[0]),
2130 2131 2132
                               qemu_opt_get(sn_opts, SNAPSHOT_OPT_ID),
                               qemu_opt_get(sn_opts, SNAPSHOT_OPT_NAME),
                               &local_err);
2133
    } else if (snapshot_name != NULL) {
P
Peter Lieven 已提交
2134
        if (s.src_num > 1) {
2135
            error_report("No support for concatenating multiple snapshot");
E
edison 已提交
2136 2137 2138
            ret = -1;
            goto out;
        }
2139

P
Peter Lieven 已提交
2140 2141
        bdrv_snapshot_load_tmp_by_id_or_name(blk_bs(s.src[0]), snapshot_name,
                                             &local_err);
2142
    }
2143
    if (local_err) {
2144
        error_reportf_err(local_err, "Failed to load snapshot: ");
2145 2146
        ret = -1;
        goto out;
E
edison 已提交
2147 2148
    }

2149
    /* Find driver and parse its options */
B
bellard 已提交
2150
    drv = bdrv_find_format(out_fmt);
2151
    if (!drv) {
2152
        error_report("Unknown file format '%s'", out_fmt);
2153 2154 2155
        ret = -1;
        goto out;
    }
2156

2157
    proto_drv = bdrv_find_protocol(out_filename, true, &local_err);
2158
    if (!proto_drv) {
2159
        error_report_err(local_err);
2160 2161 2162
        ret = -1;
        goto out;
    }
2163

M
Max Reitz 已提交
2164 2165 2166 2167 2168 2169 2170
    if (!skip_create) {
        if (!drv->create_opts) {
            error_report("Format driver '%s' does not support image creation",
                         drv->format_name);
            ret = -1;
            goto out;
        }
2171

M
Max Reitz 已提交
2172 2173 2174 2175 2176 2177
        if (!proto_drv->create_opts) {
            error_report("Protocol driver '%s' does not support image creation",
                         proto_drv->format_name);
            ret = -1;
            goto out;
        }
2178

M
Max Reitz 已提交
2179 2180
        create_opts = qemu_opts_append(create_opts, drv->create_opts);
        create_opts = qemu_opts_append(create_opts, proto_drv->create_opts);
2181

M
Max Reitz 已提交
2182
        opts = qemu_opts_create(create_opts, NULL, 0, &error_abort);
2183 2184 2185
        if (options) {
            qemu_opts_do_parse(opts, options, NULL, &local_err);
            if (local_err) {
2186
                error_report_err(local_err);
2187 2188 2189
                ret = -1;
                goto out;
            }
M
Max Reitz 已提交
2190
        }
2191

P
Peter Lieven 已提交
2192
        qemu_opt_set_number(opts, BLOCK_OPT_SIZE, s.total_sectors * 512,
2193
                            &error_abort);
M
Max Reitz 已提交
2194 2195 2196 2197
        ret = add_old_style_options(out_fmt, opts, out_baseimg, NULL);
        if (ret < 0) {
            goto out;
        }
2198
    }
2199

2200
    /* Get backing file name if -o backing_file was used */
2201
    out_baseimg_param = qemu_opt_get(opts, BLOCK_OPT_BACKING_FILE);
2202
    if (out_baseimg_param) {
2203
        out_baseimg = out_baseimg_param;
2204
    }
P
Peter Lieven 已提交
2205
    s.target_has_backing = (bool) out_baseimg;
2206

2207 2208 2209 2210 2211 2212 2213
    if (s.src_num > 1 && out_baseimg) {
        error_report("Having a backing file for the target makes no sense when "
                     "concatenating multiple input images");
        ret = -1;
        goto out;
    }

2214
    /* Check if compression is supported */
P
Peter Lieven 已提交
2215
    if (s.compressed) {
2216 2217 2218 2219
        bool encryption =
            qemu_opt_get_bool(opts, BLOCK_OPT_ENCRYPT, false);
        const char *preallocation =
            qemu_opt_get(opts, BLOCK_OPT_PREALLOC);
2220

2221
        if (!drv->bdrv_co_pwritev_compressed) {
2222
            error_report("Compression not supported for this file format");
2223 2224
            ret = -1;
            goto out;
2225 2226
        }

2227
        if (encryption) {
2228 2229
            error_report("Compression and encryption not supported at "
                         "the same time");
2230 2231
            ret = -1;
            goto out;
2232
        }
2233

2234 2235
        if (preallocation
            && strcmp(preallocation, "off"))
2236 2237 2238 2239 2240 2241
        {
            error_report("Compression and preallocation not supported at "
                         "the same time");
            ret = -1;
            goto out;
        }
2242 2243
    }

2244 2245
    if (!skip_create) {
        /* Create the new image */
C
Chunyan Liu 已提交
2246
        ret = bdrv_create(drv, out_filename, opts, &local_err);
2247
        if (ret < 0) {
2248 2249
            error_reportf_err(local_err, "%s: error while converting %s: ",
                              out_filename, out_fmt);
2250
            goto out;
B
bellard 已提交
2251 2252
        }
    }
2253

P
Peter Lieven 已提交
2254
    flags = s.min_sparse ? (BDRV_O_RDWR | BDRV_O_UNMAP) : BDRV_O_RDWR;
2255
    ret = bdrv_parse_cache_mode(cache, &flags, &writethrough);
2256 2257
    if (ret < 0) {
        error_report("Invalid cache option: %s", cache);
2258
        goto out;
2259 2260
    }

2261 2262 2263 2264 2265
    /* XXX we should allow --image-opts to trigger use of
     * img_open() here, but then we have trouble with
     * the bdrv_create() call which takes different params.
     * Not critical right now, so fix can wait...
     */
2266 2267
    s.target = img_open_file(out_filename, out_fmt, flags, writethrough, quiet,
                             false);
P
Peter Lieven 已提交
2268
    if (!s.target) {
2269 2270 2271
        ret = -1;
        goto out;
    }
P
Peter Lieven 已提交
2272
    out_bs = blk_bs(s.target);
B
bellard 已提交
2273

2274
    /* increase bufsectors from the default 4096 (2M) if opt_transfer
2275 2276
     * or discard_alignment of the out_bs is greater. Limit to 32768 (16MB)
     * as maximum. */
P
Peter Lieven 已提交
2277 2278 2279 2280 2281
    s.buf_sectors = MIN(32768,
                        MAX(s.buf_sectors,
                            MAX(out_bs->bl.opt_transfer >> BDRV_SECTOR_BITS,
                                out_bs->bl.pdiscard_alignment >>
                                BDRV_SECTOR_BITS)));
2282

2283
    if (skip_create) {
P
Peter Lieven 已提交
2284
        int64_t output_sectors = blk_nb_sectors(s.target);
2285
        if (output_sectors < 0) {
2286
            error_report("unable to get output image length: %s",
2287
                         strerror(-output_sectors));
2288 2289
            ret = -1;
            goto out;
P
Peter Lieven 已提交
2290
        } else if (output_sectors < s.total_sectors) {
2291 2292 2293 2294 2295 2296
            error_report("output file is smaller than input file");
            ret = -1;
            goto out;
        }
    }

2297 2298
    ret = bdrv_get_info(out_bs, &bdi);
    if (ret < 0) {
P
Peter Lieven 已提交
2299
        if (s.compressed) {
2300
            error_report("could not get block driver info");
2301 2302
            goto out;
        }
2303
    } else {
P
Peter Lieven 已提交
2304 2305 2306
        s.compressed = s.compressed || bdi.needs_compressed_writes;
        s.cluster_sectors = bdi.cluster_size / BDRV_SECTOR_SIZE;
    }
2307

P
Peter Lieven 已提交
2308
    ret = convert_do_copy(&s);
2309
out:
2310 2311 2312
    if (!ret) {
        qemu_progress_print(100, 0);
    }
2313
    qemu_progress_end();
2314 2315
    qemu_opts_del(opts);
    qemu_opts_free(create_opts);
2316
    qemu_opts_del(sn_opts);
P
Peter Lieven 已提交
2317 2318 2319 2320
    blk_unref(s.target);
    if (s.src) {
        for (bs_i = 0; bs_i < s.src_num; bs_i++) {
            blk_unref(s.src[bs_i]);
M
Markus Armbruster 已提交
2321
        }
P
Peter Lieven 已提交
2322
        g_free(s.src);
M
Markus Armbruster 已提交
2323
    }
P
Peter Lieven 已提交
2324
    g_free(s.src_sectors);
2325 2326 2327
fail_getopt:
    g_free(options);

P
Peter Lieven 已提交
2328
    return !!ret;
B
bellard 已提交
2329 2330
}

B
bellard 已提交
2331

B
bellard 已提交
2332 2333 2334 2335 2336 2337 2338 2339 2340
static void dump_snapshots(BlockDriverState *bs)
{
    QEMUSnapshotInfo *sn_tab, *sn;
    int nb_sns, i;

    nb_sns = bdrv_snapshot_list(bs, &sn_tab);
    if (nb_sns <= 0)
        return;
    printf("Snapshot list:\n");
2341 2342
    bdrv_snapshot_dump(fprintf, stdout, NULL);
    printf("\n");
B
bellard 已提交
2343 2344
    for(i = 0; i < nb_sns; i++) {
        sn = &sn_tab[i];
2345 2346
        bdrv_snapshot_dump(fprintf, stdout, sn);
        printf("\n");
B
bellard 已提交
2347
    }
2348
    g_free(sn_tab);
B
bellard 已提交
2349 2350
}

2351 2352 2353 2354
static void dump_json_image_info_list(ImageInfoList *list)
{
    QString *str;
    QObject *obj;
2355
    Visitor *v = qobject_output_visitor_new(&obj);
2356 2357 2358

    visit_type_ImageInfoList(v, NULL, &list, &error_abort);
    visit_complete(v, &obj);
2359 2360 2361 2362
    str = qobject_to_json_pretty(obj);
    assert(str != NULL);
    printf("%s\n", qstring_get_str(str));
    qobject_decref(obj);
2363
    visit_free(v);
2364 2365 2366
    QDECREF(str);
}

2367 2368 2369 2370
static void dump_json_image_info(ImageInfo *info)
{
    QString *str;
    QObject *obj;
2371
    Visitor *v = qobject_output_visitor_new(&obj);
2372 2373 2374

    visit_type_ImageInfo(v, NULL, &info, &error_abort);
    visit_complete(v, &obj);
2375 2376 2377 2378
    str = qobject_to_json_pretty(obj);
    assert(str != NULL);
    printf("%s\n", qstring_get_str(str));
    qobject_decref(obj);
2379
    visit_free(v);
2380 2381 2382
    QDECREF(str);
}

2383 2384 2385 2386 2387 2388 2389 2390 2391 2392 2393
static void dump_human_image_info_list(ImageInfoList *list)
{
    ImageInfoList *elem;
    bool delim = false;

    for (elem = list; elem; elem = elem->next) {
        if (delim) {
            printf("\n");
        }
        delim = true;

2394
        bdrv_image_info_dump(fprintf, stdout, elem->value);
2395 2396 2397 2398 2399 2400 2401 2402 2403 2404 2405 2406 2407 2408 2409 2410 2411 2412 2413 2414
    }
}

static gboolean str_equal_func(gconstpointer a, gconstpointer b)
{
    return strcmp(a, b) == 0;
}

/**
 * Open an image file chain and return an ImageInfoList
 *
 * @filename: topmost image filename
 * @fmt: topmost image format (may be NULL to autodetect)
 * @chain: true  - enumerate entire backing file chain
 *         false - only topmost image file
 *
 * Returns a list of ImageInfo objects or NULL if there was an error opening an
 * image file.  If there was an error a message will have been printed to
 * stderr.
 */
2415 2416
static ImageInfoList *collect_image_info_list(bool image_opts,
                                              const char *filename,
2417
                                              const char *fmt,
2418
                                              bool chain, bool force_share)
2419 2420 2421 2422
{
    ImageInfoList *head = NULL;
    ImageInfoList **last = &head;
    GHashTable *filenames;
2423
    Error *err = NULL;
2424 2425 2426 2427

    filenames = g_hash_table_new_full(g_str_hash, str_equal_func, NULL, NULL);

    while (filename) {
M
Markus Armbruster 已提交
2428
        BlockBackend *blk;
2429 2430 2431 2432 2433 2434 2435 2436 2437 2438 2439
        BlockDriverState *bs;
        ImageInfo *info;
        ImageInfoList *elem;

        if (g_hash_table_lookup_extended(filenames, filename, NULL, NULL)) {
            error_report("Backing file '%s' creates an infinite loop.",
                         filename);
            goto err;
        }
        g_hash_table_insert(filenames, (gpointer)filename, NULL);

2440
        blk = img_open(image_opts, filename, fmt,
2441 2442
                       BDRV_O_NO_BACKING | BDRV_O_NO_IO, false, false,
                       force_share);
2443
        if (!blk) {
2444 2445
            goto err;
        }
2446
        bs = blk_bs(blk);
2447

2448
        bdrv_query_image_info(bs, &info, &err);
2449
        if (err) {
2450
            error_report_err(err);
M
Markus Armbruster 已提交
2451
            blk_unref(blk);
2452
            goto err;
2453
        }
2454 2455 2456 2457 2458 2459

        elem = g_new0(ImageInfoList, 1);
        elem->value = info;
        *last = elem;
        last = &elem->next;

M
Markus Armbruster 已提交
2460
        blk_unref(blk);
2461 2462 2463 2464 2465 2466

        filename = fmt = NULL;
        if (chain) {
            if (info->has_full_backing_filename) {
                filename = info->full_backing_filename;
            } else if (info->has_backing_filename) {
2467 2468 2469 2470
                error_report("Could not determine absolute backing filename,"
                             " but backing filename '%s' present",
                             info->backing_filename);
                goto err;
2471 2472 2473 2474 2475 2476 2477 2478 2479 2480 2481 2482 2483 2484 2485
            }
            if (info->has_backing_filename_format) {
                fmt = info->backing_filename_format;
            }
        }
    }
    g_hash_table_destroy(filenames);
    return head;

err:
    qapi_free_ImageInfoList(head);
    g_hash_table_destroy(filenames);
    return NULL;
}

2486 2487 2488 2489
static int img_info(int argc, char **argv)
{
    int c;
    OutputFormat output_format = OFORMAT_HUMAN;
2490
    bool chain = false;
2491
    const char *filename, *fmt, *output;
2492
    ImageInfoList *list;
2493
    bool image_opts = false;
2494
    bool force_share = false;
2495

B
bellard 已提交
2496
    fmt = NULL;
2497
    output = NULL;
B
bellard 已提交
2498
    for(;;) {
2499 2500 2501 2502 2503
        int option_index = 0;
        static const struct option long_options[] = {
            {"help", no_argument, 0, 'h'},
            {"format", required_argument, 0, 'f'},
            {"output", required_argument, 0, OPTION_OUTPUT},
2504
            {"backing-chain", no_argument, 0, OPTION_BACKING_CHAIN},
2505
            {"object", required_argument, 0, OPTION_OBJECT},
2506
            {"image-opts", no_argument, 0, OPTION_IMAGE_OPTS},
2507
            {"force-share", no_argument, 0, 'U'},
2508 2509
            {0, 0, 0, 0}
        };
2510
        c = getopt_long(argc, argv, ":f:hU",
2511
                        long_options, &option_index);
2512
        if (c == -1) {
B
bellard 已提交
2513
            break;
2514
        }
B
bellard 已提交
2515
        switch(c) {
2516 2517 2518
        case ':':
            missing_argument(argv[optind - 1]);
            break;
J
Jes Sorensen 已提交
2519
        case '?':
2520 2521
            unrecognized_option(argv[optind - 1]);
            break;
B
bellard 已提交
2522 2523 2524 2525 2526 2527
        case 'h':
            help();
            break;
        case 'f':
            fmt = optarg;
            break;
2528 2529 2530
        case 'U':
            force_share = true;
            break;
2531 2532 2533
        case OPTION_OUTPUT:
            output = optarg;
            break;
2534 2535 2536
        case OPTION_BACKING_CHAIN:
            chain = true;
            break;
2537 2538 2539 2540 2541 2542 2543 2544
        case OPTION_OBJECT: {
            QemuOpts *opts;
            opts = qemu_opts_parse_noisily(&qemu_object_opts,
                                           optarg, true);
            if (!opts) {
                return 1;
            }
        }   break;
2545 2546 2547
        case OPTION_IMAGE_OPTS:
            image_opts = true;
            break;
B
bellard 已提交
2548 2549
        }
    }
2550
    if (optind != argc - 1) {
F
Fam Zheng 已提交
2551
        error_exit("Expecting one image file name");
2552
    }
B
bellard 已提交
2553 2554
    filename = argv[optind++];

2555 2556 2557 2558 2559 2560
    if (output && !strcmp(output, "json")) {
        output_format = OFORMAT_JSON;
    } else if (output && !strcmp(output, "human")) {
        output_format = OFORMAT_HUMAN;
    } else if (output) {
        error_report("--output must be used with human or json as argument.");
2561 2562
        return 1;
    }
2563

2564 2565
    if (qemu_opts_foreach(&qemu_object_opts,
                          user_creatable_add_opts_foreach,
2566
                          NULL, NULL)) {
2567 2568 2569
        return 1;
    }

2570 2571
    list = collect_image_info_list(image_opts, filename, fmt, chain,
                                   force_share);
2572
    if (!list) {
2573
        return 1;
B
bellard 已提交
2574
    }
2575 2576 2577

    switch (output_format) {
    case OFORMAT_HUMAN:
2578
        dump_human_image_info_list(list);
2579 2580
        break;
    case OFORMAT_JSON:
2581 2582 2583 2584 2585
        if (chain) {
            dump_json_image_info_list(list);
        } else {
            dump_json_image_info(list->value);
        }
2586
        break;
B
bellard 已提交
2587
    }
2588

2589
    qapi_free_ImageInfoList(list);
B
bellard 已提交
2590 2591 2592
    return 0;
}

P
Paolo Bonzini 已提交
2593 2594 2595 2596 2597
static void dump_map_entry(OutputFormat output_format, MapEntry *e,
                           MapEntry *next)
{
    switch (output_format) {
    case OFORMAT_HUMAN:
2598
        if (e->data && !e->has_offset) {
P
Paolo Bonzini 已提交
2599 2600 2601
            error_report("File contains external, encrypted or compressed clusters.");
            exit(1);
        }
2602
        if (e->data && !e->zero) {
P
Paolo Bonzini 已提交
2603
            printf("%#-16"PRIx64"%#-16"PRIx64"%#-16"PRIx64"%s\n",
2604 2605 2606
                   e->start, e->length,
                   e->has_offset ? e->offset : 0,
                   e->has_filename ? e->filename : "");
P
Paolo Bonzini 已提交
2607 2608 2609 2610
        }
        /* This format ignores the distinction between 0, ZERO and ZERO|DATA.
         * Modify the flags here to allow more coalescing.
         */
2611 2612 2613
        if (next && (!next->data || next->zero)) {
            next->data = false;
            next->zero = true;
P
Paolo Bonzini 已提交
2614 2615 2616
        }
        break;
    case OFORMAT_JSON:
2617 2618
        printf("%s{ \"start\": %"PRId64", \"length\": %"PRId64","
               " \"depth\": %"PRId64", \"zero\": %s, \"data\": %s",
P
Paolo Bonzini 已提交
2619 2620
               (e->start == 0 ? "[" : ",\n"),
               e->start, e->length, e->depth,
2621 2622 2623
               e->zero ? "true" : "false",
               e->data ? "true" : "false");
        if (e->has_offset) {
P
Paolo Bonzini 已提交
2624
            printf(", \"offset\": %"PRId64"", e->offset);
P
Paolo Bonzini 已提交
2625 2626 2627 2628 2629 2630 2631 2632 2633 2634 2635 2636 2637 2638 2639
        }
        putchar('}');

        if (!next) {
            printf("]\n");
        }
        break;
    }
}

static int get_block_status(BlockDriverState *bs, int64_t sector_num,
                            int nb_sectors, MapEntry *e)
{
    int64_t ret;
    int depth;
2640
    BlockDriverState *file;
J
John Snow 已提交
2641
    bool has_offset;
P
Paolo Bonzini 已提交
2642 2643 2644 2645 2646 2647 2648 2649

    /* As an optimization, we could cache the current range of unallocated
     * clusters in each file of the chain, and avoid querying the same
     * range repeatedly.
     */

    depth = 0;
    for (;;) {
2650 2651
        ret = bdrv_get_block_status(bs, sector_num, nb_sectors, &nb_sectors,
                                    &file);
P
Paolo Bonzini 已提交
2652 2653 2654 2655 2656 2657 2658
        if (ret < 0) {
            return ret;
        }
        assert(nb_sectors);
        if (ret & (BDRV_BLOCK_ZERO|BDRV_BLOCK_DATA)) {
            break;
        }
2659
        bs = backing_bs(bs);
P
Paolo Bonzini 已提交
2660 2661 2662 2663 2664 2665 2666 2667
        if (bs == NULL) {
            ret = 0;
            break;
        }

        depth++;
    }

J
John Snow 已提交
2668 2669 2670 2671 2672 2673 2674 2675 2676 2677 2678 2679 2680 2681
    has_offset = !!(ret & BDRV_BLOCK_OFFSET_VALID);

    *e = (MapEntry) {
        .start = sector_num * BDRV_SECTOR_SIZE,
        .length = nb_sectors * BDRV_SECTOR_SIZE,
        .data = !!(ret & BDRV_BLOCK_DATA),
        .zero = !!(ret & BDRV_BLOCK_ZERO),
        .offset = ret & BDRV_BLOCK_OFFSET_MASK,
        .has_offset = has_offset,
        .depth = depth,
        .has_filename = file && has_offset,
        .filename = file && has_offset ? file->filename : NULL,
    };

P
Paolo Bonzini 已提交
2682 2683 2684
    return 0;
}

2685 2686 2687 2688 2689 2690 2691 2692 2693 2694 2695 2696 2697 2698 2699 2700 2701 2702 2703 2704 2705
static inline bool entry_mergeable(const MapEntry *curr, const MapEntry *next)
{
    if (curr->length == 0) {
        return false;
    }
    if (curr->zero != next->zero ||
        curr->data != next->data ||
        curr->depth != next->depth ||
        curr->has_filename != next->has_filename ||
        curr->has_offset != next->has_offset) {
        return false;
    }
    if (curr->has_filename && strcmp(curr->filename, next->filename)) {
        return false;
    }
    if (curr->has_offset && curr->offset + curr->length != next->offset) {
        return false;
    }
    return true;
}

P
Paolo Bonzini 已提交
2706 2707 2708 2709
static int img_map(int argc, char **argv)
{
    int c;
    OutputFormat output_format = OFORMAT_HUMAN;
M
Markus Armbruster 已提交
2710
    BlockBackend *blk;
P
Paolo Bonzini 已提交
2711 2712 2713 2714 2715
    BlockDriverState *bs;
    const char *filename, *fmt, *output;
    int64_t length;
    MapEntry curr = { .length = 0 }, next;
    int ret = 0;
2716
    bool image_opts = false;
2717
    bool force_share = false;
P
Paolo Bonzini 已提交
2718 2719 2720 2721 2722 2723 2724 2725 2726

    fmt = NULL;
    output = NULL;
    for (;;) {
        int option_index = 0;
        static const struct option long_options[] = {
            {"help", no_argument, 0, 'h'},
            {"format", required_argument, 0, 'f'},
            {"output", required_argument, 0, OPTION_OUTPUT},
2727
            {"object", required_argument, 0, OPTION_OBJECT},
2728
            {"image-opts", no_argument, 0, OPTION_IMAGE_OPTS},
2729
            {"force-share", no_argument, 0, 'U'},
P
Paolo Bonzini 已提交
2730 2731
            {0, 0, 0, 0}
        };
2732
        c = getopt_long(argc, argv, ":f:hU",
P
Paolo Bonzini 已提交
2733 2734 2735 2736 2737
                        long_options, &option_index);
        if (c == -1) {
            break;
        }
        switch (c) {
2738 2739 2740
        case ':':
            missing_argument(argv[optind - 1]);
            break;
P
Paolo Bonzini 已提交
2741
        case '?':
2742 2743
            unrecognized_option(argv[optind - 1]);
            break;
P
Paolo Bonzini 已提交
2744 2745 2746 2747 2748 2749
        case 'h':
            help();
            break;
        case 'f':
            fmt = optarg;
            break;
2750 2751 2752
        case 'U':
            force_share = true;
            break;
P
Paolo Bonzini 已提交
2753 2754 2755
        case OPTION_OUTPUT:
            output = optarg;
            break;
2756 2757 2758 2759 2760 2761 2762 2763
        case OPTION_OBJECT: {
            QemuOpts *opts;
            opts = qemu_opts_parse_noisily(&qemu_object_opts,
                                           optarg, true);
            if (!opts) {
                return 1;
            }
        }   break;
2764 2765 2766
        case OPTION_IMAGE_OPTS:
            image_opts = true;
            break;
P
Paolo Bonzini 已提交
2767 2768
        }
    }
F
Fam Zheng 已提交
2769 2770
    if (optind != argc - 1) {
        error_exit("Expecting one image file name");
P
Paolo Bonzini 已提交
2771
    }
F
Fam Zheng 已提交
2772
    filename = argv[optind];
P
Paolo Bonzini 已提交
2773 2774 2775 2776 2777 2778 2779 2780 2781 2782

    if (output && !strcmp(output, "json")) {
        output_format = OFORMAT_JSON;
    } else if (output && !strcmp(output, "human")) {
        output_format = OFORMAT_HUMAN;
    } else if (output) {
        error_report("--output must be used with human or json as argument.");
        return 1;
    }

2783 2784
    if (qemu_opts_foreach(&qemu_object_opts,
                          user_creatable_add_opts_foreach,
2785
                          NULL, NULL)) {
2786 2787 2788
        return 1;
    }

2789
    blk = img_open(image_opts, filename, fmt, 0, false, false, force_share);
2790 2791
    if (!blk) {
        return 1;
P
Paolo Bonzini 已提交
2792
    }
2793
    bs = blk_bs(blk);
P
Paolo Bonzini 已提交
2794 2795 2796 2797 2798

    if (output_format == OFORMAT_HUMAN) {
        printf("%-16s%-16s%-16s%s\n", "Offset", "Length", "Mapped to", "File");
    }

2799
    length = blk_getlength(blk);
P
Paolo Bonzini 已提交
2800 2801 2802 2803 2804 2805 2806 2807 2808 2809 2810 2811 2812 2813 2814 2815 2816
    while (curr.start + curr.length < length) {
        int64_t nsectors_left;
        int64_t sector_num;
        int n;

        sector_num = (curr.start + curr.length) >> BDRV_SECTOR_BITS;

        /* Probe up to 1 GiB at a time.  */
        nsectors_left = DIV_ROUND_UP(length, BDRV_SECTOR_SIZE) - sector_num;
        n = MIN(1 << (30 - BDRV_SECTOR_BITS), nsectors_left);
        ret = get_block_status(bs, sector_num, n, &next);

        if (ret < 0) {
            error_report("Could not read file metadata: %s", strerror(-ret));
            goto out;
        }

2817
        if (entry_mergeable(&curr, &next)) {
P
Paolo Bonzini 已提交
2818 2819 2820 2821 2822 2823 2824 2825 2826 2827 2828 2829 2830
            curr.length += next.length;
            continue;
        }

        if (curr.length > 0) {
            dump_map_entry(output_format, &curr, &next);
        }
        curr = next;
    }

    dump_map_entry(output_format, &curr, NULL);

out:
M
Markus Armbruster 已提交
2831
    blk_unref(blk);
P
Paolo Bonzini 已提交
2832 2833 2834
    return ret < 0;
}

2835 2836 2837 2838 2839
#define SNAPSHOT_LIST   1
#define SNAPSHOT_CREATE 2
#define SNAPSHOT_APPLY  3
#define SNAPSHOT_DELETE 4

2840
static int img_snapshot(int argc, char **argv)
2841
{
M
Markus Armbruster 已提交
2842
    BlockBackend *blk;
2843 2844 2845
    BlockDriverState *bs;
    QEMUSnapshotInfo sn;
    char *filename, *snapshot_name = NULL;
2846
    int c, ret = 0, bdrv_oflags;
2847 2848
    int action = 0;
    qemu_timeval tv;
2849
    bool quiet = false;
2850
    Error *err = NULL;
2851
    bool image_opts = false;
2852
    bool force_share = false;
2853

2854
    bdrv_oflags = BDRV_O_RDWR;
2855 2856
    /* Parse commandline parameters */
    for(;;) {
2857 2858 2859
        static const struct option long_options[] = {
            {"help", no_argument, 0, 'h'},
            {"object", required_argument, 0, OPTION_OBJECT},
2860
            {"image-opts", no_argument, 0, OPTION_IMAGE_OPTS},
2861
            {"force-share", no_argument, 0, 'U'},
2862 2863
            {0, 0, 0, 0}
        };
2864
        c = getopt_long(argc, argv, ":la:c:d:hqU",
2865
                        long_options, NULL);
2866
        if (c == -1) {
2867
            break;
2868
        }
2869
        switch(c) {
2870 2871 2872
        case ':':
            missing_argument(argv[optind - 1]);
            break;
J
Jes Sorensen 已提交
2873
        case '?':
2874 2875
            unrecognized_option(argv[optind - 1]);
            break;
2876 2877
        case 'h':
            help();
2878
            return 0;
2879 2880
        case 'l':
            if (action) {
F
Fam Zheng 已提交
2881
                error_exit("Cannot mix '-l', '-a', '-c', '-d'");
2882
                return 0;
2883 2884
            }
            action = SNAPSHOT_LIST;
2885
            bdrv_oflags &= ~BDRV_O_RDWR; /* no need for RW */
2886 2887 2888
            break;
        case 'a':
            if (action) {
F
Fam Zheng 已提交
2889
                error_exit("Cannot mix '-l', '-a', '-c', '-d'");
2890
                return 0;
2891 2892 2893 2894 2895 2896
            }
            action = SNAPSHOT_APPLY;
            snapshot_name = optarg;
            break;
        case 'c':
            if (action) {
F
Fam Zheng 已提交
2897
                error_exit("Cannot mix '-l', '-a', '-c', '-d'");
2898
                return 0;
2899 2900 2901 2902 2903 2904
            }
            action = SNAPSHOT_CREATE;
            snapshot_name = optarg;
            break;
        case 'd':
            if (action) {
F
Fam Zheng 已提交
2905
                error_exit("Cannot mix '-l', '-a', '-c', '-d'");
2906
                return 0;
2907 2908 2909 2910
            }
            action = SNAPSHOT_DELETE;
            snapshot_name = optarg;
            break;
2911 2912 2913
        case 'q':
            quiet = true;
            break;
2914 2915 2916
        case 'U':
            force_share = true;
            break;
2917 2918 2919 2920 2921 2922 2923 2924
        case OPTION_OBJECT: {
            QemuOpts *opts;
            opts = qemu_opts_parse_noisily(&qemu_object_opts,
                                           optarg, true);
            if (!opts) {
                return 1;
            }
        }   break;
2925 2926 2927
        case OPTION_IMAGE_OPTS:
            image_opts = true;
            break;
2928 2929 2930
        }
    }

2931
    if (optind != argc - 1) {
F
Fam Zheng 已提交
2932
        error_exit("Expecting one image file name");
2933
    }
2934 2935
    filename = argv[optind++];

2936 2937
    if (qemu_opts_foreach(&qemu_object_opts,
                          user_creatable_add_opts_foreach,
2938
                          NULL, NULL)) {
2939 2940 2941
        return 1;
    }

2942
    /* Open the image */
2943 2944
    blk = img_open(image_opts, filename, NULL, bdrv_oflags, false, quiet,
                   force_share);
2945 2946
    if (!blk) {
        return 1;
2947
    }
2948
    bs = blk_bs(blk);
2949 2950 2951 2952 2953 2954 2955 2956 2957 2958 2959 2960 2961 2962 2963 2964

    /* Perform the requested action */
    switch(action) {
    case SNAPSHOT_LIST:
        dump_snapshots(bs);
        break;

    case SNAPSHOT_CREATE:
        memset(&sn, 0, sizeof(sn));
        pstrcpy(sn.name, sizeof(sn.name), snapshot_name);

        qemu_gettimeofday(&tv);
        sn.date_sec = tv.tv_sec;
        sn.date_nsec = tv.tv_usec * 1000;

        ret = bdrv_snapshot_create(bs, &sn);
2965
        if (ret) {
2966
            error_report("Could not create snapshot '%s': %d (%s)",
2967
                snapshot_name, ret, strerror(-ret));
2968
        }
2969 2970 2971 2972
        break;

    case SNAPSHOT_APPLY:
        ret = bdrv_snapshot_goto(bs, snapshot_name);
2973
        if (ret) {
2974
            error_report("Could not apply snapshot '%s': %d (%s)",
2975
                snapshot_name, ret, strerror(-ret));
2976
        }
2977 2978 2979
        break;

    case SNAPSHOT_DELETE:
2980
        bdrv_snapshot_delete_by_id_or_name(bs, snapshot_name, &err);
2981
        if (err) {
2982 2983
            error_reportf_err(err, "Could not delete snapshot '%s': ",
                              snapshot_name);
2984
            ret = 1;
2985
        }
2986 2987 2988 2989
        break;
    }

    /* Cleanup */
M
Markus Armbruster 已提交
2990
    blk_unref(blk);
2991 2992 2993
    if (ret) {
        return 1;
    }
2994
    return 0;
2995 2996
}

K
Kevin Wolf 已提交
2997 2998
static int img_rebase(int argc, char **argv)
{
M
Markus Armbruster 已提交
2999
    BlockBackend *blk = NULL, *blk_old_backing = NULL, *blk_new_backing = NULL;
P
Paolo Bonzini 已提交
3000 3001
    uint8_t *buf_old = NULL;
    uint8_t *buf_new = NULL;
3002
    BlockDriverState *bs = NULL;
K
Kevin Wolf 已提交
3003
    char *filename;
3004 3005
    const char *fmt, *cache, *src_cache, *out_basefmt, *out_baseimg;
    int c, flags, src_flags, ret;
3006
    bool writethrough, src_writethrough;
K
Kevin Wolf 已提交
3007
    int unsafe = 0;
3008
    bool force_share = false;
3009
    int progress = 0;
3010
    bool quiet = false;
3011
    Error *local_err = NULL;
3012
    bool image_opts = false;
K
Kevin Wolf 已提交
3013 3014

    /* Parse commandline parameters */
K
Kevin Wolf 已提交
3015
    fmt = NULL;
3016
    cache = BDRV_DEFAULT_CACHE;
3017
    src_cache = BDRV_DEFAULT_CACHE;
K
Kevin Wolf 已提交
3018 3019 3020
    out_baseimg = NULL;
    out_basefmt = NULL;
    for(;;) {
3021 3022 3023
        static const struct option long_options[] = {
            {"help", no_argument, 0, 'h'},
            {"object", required_argument, 0, OPTION_OBJECT},
3024
            {"image-opts", no_argument, 0, OPTION_IMAGE_OPTS},
3025
            {"force-share", no_argument, 0, 'U'},
3026 3027
            {0, 0, 0, 0}
        };
3028
        c = getopt_long(argc, argv, ":hf:F:b:upt:T:qU",
3029
                        long_options, NULL);
3030
        if (c == -1) {
K
Kevin Wolf 已提交
3031
            break;
3032
        }
K
Kevin Wolf 已提交
3033
        switch(c) {
3034 3035 3036
        case ':':
            missing_argument(argv[optind - 1]);
            break;
J
Jes Sorensen 已提交
3037
        case '?':
3038 3039
            unrecognized_option(argv[optind - 1]);
            break;
K
Kevin Wolf 已提交
3040 3041 3042
        case 'h':
            help();
            return 0;
K
Kevin Wolf 已提交
3043 3044 3045
        case 'f':
            fmt = optarg;
            break;
K
Kevin Wolf 已提交
3046 3047 3048 3049 3050 3051 3052 3053 3054
        case 'F':
            out_basefmt = optarg;
            break;
        case 'b':
            out_baseimg = optarg;
            break;
        case 'u':
            unsafe = 1;
            break;
3055 3056 3057
        case 'p':
            progress = 1;
            break;
3058 3059 3060
        case 't':
            cache = optarg;
            break;
3061 3062 3063
        case 'T':
            src_cache = optarg;
            break;
3064 3065 3066
        case 'q':
            quiet = true;
            break;
3067 3068 3069 3070 3071 3072 3073 3074
        case OPTION_OBJECT: {
            QemuOpts *opts;
            opts = qemu_opts_parse_noisily(&qemu_object_opts,
                                           optarg, true);
            if (!opts) {
                return 1;
            }
        }   break;
3075 3076 3077
        case OPTION_IMAGE_OPTS:
            image_opts = true;
            break;
3078 3079 3080
        case 'U':
            force_share = true;
            break;
K
Kevin Wolf 已提交
3081 3082 3083
        }
    }

3084 3085 3086 3087
    if (quiet) {
        progress = 0;
    }

F
Fam Zheng 已提交
3088 3089 3090 3091 3092
    if (optind != argc - 1) {
        error_exit("Expecting one image file name");
    }
    if (!unsafe && !out_baseimg) {
        error_exit("Must specify backing file (-b) or use unsafe mode (-u)");
3093
    }
K
Kevin Wolf 已提交
3094 3095
    filename = argv[optind++];

3096 3097
    if (qemu_opts_foreach(&qemu_object_opts,
                          user_creatable_add_opts_foreach,
3098
                          NULL, NULL)) {
3099 3100 3101
        return 1;
    }

3102 3103 3104
    qemu_progress_init(progress, 2.0);
    qemu_progress_print(0, 100);

3105
    flags = BDRV_O_RDWR | (unsafe ? BDRV_O_NO_BACKING : 0);
3106
    ret = bdrv_parse_cache_mode(cache, &flags, &writethrough);
3107 3108
    if (ret < 0) {
        error_report("Invalid cache option: %s", cache);
3109
        goto out;
3110 3111
    }

3112 3113
    src_flags = 0;
    ret = bdrv_parse_cache_mode(src_cache, &src_flags, &src_writethrough);
3114 3115
    if (ret < 0) {
        error_report("Invalid source cache option: %s", src_cache);
3116
        goto out;
3117 3118
    }

3119 3120 3121 3122
    /* The source files are opened read-only, don't care about WCE */
    assert((src_flags & BDRV_O_RDWR) == 0);
    (void) src_writethrough;

K
Kevin Wolf 已提交
3123 3124 3125 3126 3127 3128
    /*
     * Open the images.
     *
     * Ignore the old backing file for unsafe rebase in case we want to correct
     * the reference to a renamed or moved backing file.
     */
3129 3130
    blk = img_open(image_opts, filename, fmt, flags, writethrough, quiet,
                   false);
3131
    if (!blk) {
3132 3133
        ret = -1;
        goto out;
3134
    }
3135
    bs = blk_bs(blk);
K
Kevin Wolf 已提交
3136 3137

    if (out_basefmt != NULL) {
3138
        if (bdrv_find_format(out_basefmt) == NULL) {
3139
            error_report("Invalid format name: '%s'", out_basefmt);
3140 3141
            ret = -1;
            goto out;
K
Kevin Wolf 已提交
3142 3143 3144 3145
        }
    }

    /* For safe rebasing we need to compare old and new backing file */
3146
    if (!unsafe) {
3147
        char backing_name[PATH_MAX];
3148 3149 3150 3151
        QDict *options = NULL;

        if (bs->backing_format[0] != '\0') {
            options = qdict_new();
3152
            qdict_put_str(options, "driver", bs->backing_format);
3153
        }
K
Kevin Wolf 已提交
3154

3155 3156 3157 3158 3159 3160 3161
        if (force_share) {
            if (!options) {
                options = qdict_new();
            }
            qdict_put(options, BDRV_OPT_FORCE_SHARE,
                      qbool_from_bool(true));
        }
K
Kevin Wolf 已提交
3162
        bdrv_get_backing_filename(bs, backing_name, sizeof(backing_name));
3163
        blk_old_backing = blk_new_open(backing_name, NULL,
3164 3165
                                       options, src_flags, &local_err);
        if (!blk_old_backing) {
3166 3167 3168
            error_reportf_err(local_err,
                              "Could not open old backing file '%s': ",
                              backing_name);
3169
            ret = -1;
3170
            goto out;
K
Kevin Wolf 已提交
3171
        }
3172

3173
        if (out_baseimg[0]) {
3174
            options = qdict_new();
3175
            if (out_basefmt) {
3176
                qdict_put_str(options, "driver", out_basefmt);
3177 3178 3179
            }
            if (force_share) {
                qdict_put_bool(options, BDRV_OPT_FORCE_SHARE, true);
3180 3181
            }

3182
            blk_new_backing = blk_new_open(out_baseimg, NULL,
3183 3184
                                           options, src_flags, &local_err);
            if (!blk_new_backing) {
3185 3186 3187
                error_reportf_err(local_err,
                                  "Could not open new backing file '%s': ",
                                  out_baseimg);
3188
                ret = -1;
3189 3190
                goto out;
            }
K
Kevin Wolf 已提交
3191 3192 3193 3194 3195 3196 3197 3198 3199 3200 3201 3202 3203
        }
    }

    /*
     * Check each unallocated cluster in the COW file. If it is unallocated,
     * accesses go to the backing file. We must therefore compare this cluster
     * in the old and new backing file, and if they differ we need to copy it
     * from the old backing file into the COW file.
     *
     * If qemu-img crashes during this step, no harm is done. The content of
     * the image is the same as the original one at any time.
     */
    if (!unsafe) {
3204 3205 3206
        int64_t num_sectors;
        int64_t old_backing_num_sectors;
        int64_t new_backing_num_sectors = 0;
K
Kevin Wolf 已提交
3207
        uint64_t sector;
3208
        int n;
3209
        float local_progress = 0;
3210

3211 3212
        buf_old = blk_blockalign(blk, IO_BUF_SIZE);
        buf_new = blk_blockalign(blk, IO_BUF_SIZE);
K
Kevin Wolf 已提交
3213

3214
        num_sectors = blk_nb_sectors(blk);
3215 3216 3217 3218 3219 3220
        if (num_sectors < 0) {
            error_report("Could not get size of '%s': %s",
                         filename, strerror(-num_sectors));
            ret = -1;
            goto out;
        }
3221
        old_backing_num_sectors = blk_nb_sectors(blk_old_backing);
3222
        if (old_backing_num_sectors < 0) {
3223
            char backing_name[PATH_MAX];
3224 3225 3226 3227 3228 3229 3230

            bdrv_get_backing_filename(bs, backing_name, sizeof(backing_name));
            error_report("Could not get size of '%s': %s",
                         backing_name, strerror(-old_backing_num_sectors));
            ret = -1;
            goto out;
        }
3231 3232
        if (blk_new_backing) {
            new_backing_num_sectors = blk_nb_sectors(blk_new_backing);
3233 3234 3235 3236 3237 3238
            if (new_backing_num_sectors < 0) {
                error_report("Could not get size of '%s': %s",
                             out_baseimg, strerror(-new_backing_num_sectors));
                ret = -1;
                goto out;
            }
3239
        }
K
Kevin Wolf 已提交
3240

3241 3242 3243 3244 3245
        if (num_sectors != 0) {
            local_progress = (float)100 /
                (num_sectors / MIN(num_sectors, IO_BUF_SIZE / 512));
        }

K
Kevin Wolf 已提交
3246 3247 3248 3249 3250 3251 3252 3253 3254 3255
        for (sector = 0; sector < num_sectors; sector += n) {

            /* How many sectors can we handle with the next read? */
            if (sector + (IO_BUF_SIZE / 512) <= num_sectors) {
                n = (IO_BUF_SIZE / 512);
            } else {
                n = num_sectors - sector;
            }

            /* If the cluster is allocated, we don't need to take action */
3256
            ret = bdrv_is_allocated(bs, sector, n, &n);
3257 3258 3259 3260 3261
            if (ret < 0) {
                error_report("error while reading image metadata: %s",
                             strerror(-ret));
                goto out;
            }
3262
            if (ret) {
K
Kevin Wolf 已提交
3263 3264 3265
                continue;
            }

3266 3267 3268 3269 3270 3271 3272 3273 3274 3275 3276
            /*
             * Read old and new backing file and take into consideration that
             * backing files may be smaller than the COW image.
             */
            if (sector >= old_backing_num_sectors) {
                memset(buf_old, 0, n * BDRV_SECTOR_SIZE);
            } else {
                if (sector + n > old_backing_num_sectors) {
                    n = old_backing_num_sectors - sector;
                }

3277 3278
                ret = blk_pread(blk_old_backing, sector << BDRV_SECTOR_BITS,
                                buf_old, n << BDRV_SECTOR_BITS);
3279 3280 3281 3282
                if (ret < 0) {
                    error_report("error while reading from old backing file");
                    goto out;
                }
K
Kevin Wolf 已提交
3283
            }
3284

3285
            if (sector >= new_backing_num_sectors || !blk_new_backing) {
3286 3287 3288 3289 3290 3291
                memset(buf_new, 0, n * BDRV_SECTOR_SIZE);
            } else {
                if (sector + n > new_backing_num_sectors) {
                    n = new_backing_num_sectors - sector;
                }

3292 3293
                ret = blk_pread(blk_new_backing, sector << BDRV_SECTOR_BITS,
                                buf_new, n << BDRV_SECTOR_BITS);
3294 3295 3296 3297
                if (ret < 0) {
                    error_report("error while reading from new backing file");
                    goto out;
                }
K
Kevin Wolf 已提交
3298 3299 3300 3301 3302 3303 3304 3305 3306
            }

            /* If they differ, we need to write to the COW file */
            uint64_t written = 0;

            while (written < n) {
                int pnum;

                if (compare_sectors(buf_old + written * 512,
3307
                    buf_new + written * 512, n - written, &pnum))
K
Kevin Wolf 已提交
3308
                {
3309 3310 3311 3312
                    ret = blk_pwrite(blk,
                                     (sector + written) << BDRV_SECTOR_BITS,
                                     buf_old + written * 512,
                                     pnum << BDRV_SECTOR_BITS, 0);
K
Kevin Wolf 已提交
3313
                    if (ret < 0) {
3314
                        error_report("Error while writing to COW image: %s",
K
Kevin Wolf 已提交
3315
                            strerror(-ret));
3316
                        goto out;
K
Kevin Wolf 已提交
3317 3318 3319 3320 3321
                    }
                }

                written += pnum;
            }
3322
            qemu_progress_print(local_progress, 100);
K
Kevin Wolf 已提交
3323 3324 3325 3326 3327 3328 3329 3330
        }
    }

    /*
     * Change the backing file. All clusters that are different from the old
     * backing file are overwritten in the COW file now, so the visible content
     * doesn't change when we switch the backing file.
     */
3331 3332 3333 3334 3335 3336
    if (out_baseimg && *out_baseimg) {
        ret = bdrv_change_backing_file(bs, out_baseimg, out_basefmt);
    } else {
        ret = bdrv_change_backing_file(bs, NULL, NULL);
    }

K
Kevin Wolf 已提交
3337
    if (ret == -ENOSPC) {
3338 3339
        error_report("Could not change the backing file to '%s': No "
                     "space left in the file header", out_baseimg);
K
Kevin Wolf 已提交
3340
    } else if (ret < 0) {
3341
        error_report("Could not change the backing file to '%s': %s",
K
Kevin Wolf 已提交
3342 3343 3344
            out_baseimg, strerror(-ret));
    }

3345
    qemu_progress_print(100, 0);
K
Kevin Wolf 已提交
3346 3347 3348 3349 3350 3351
    /*
     * TODO At this point it is possible to check if any clusters that are
     * allocated in the COW file are the same in the backing file. If so, they
     * could be dropped from the COW file. Don't do this before switching the
     * backing file, in case of a crash this would lead to corruption.
     */
3352
out:
3353
    qemu_progress_end();
K
Kevin Wolf 已提交
3354 3355
    /* Cleanup */
    if (!unsafe) {
M
Markus Armbruster 已提交
3356 3357
        blk_unref(blk_old_backing);
        blk_unref(blk_new_backing);
K
Kevin Wolf 已提交
3358
    }
P
Paolo Bonzini 已提交
3359 3360
    qemu_vfree(buf_old);
    qemu_vfree(buf_new);
K
Kevin Wolf 已提交
3361

M
Markus Armbruster 已提交
3362
    blk_unref(blk);
3363 3364 3365
    if (ret) {
        return 1;
    }
K
Kevin Wolf 已提交
3366 3367 3368
    return 0;
}

3369 3370
static int img_resize(int argc, char **argv)
{
3371
    Error *err = NULL;
3372 3373 3374
    int c, ret, relative;
    const char *filename, *fmt, *size;
    int64_t n, total_size;
3375
    bool quiet = false;
M
Markus Armbruster 已提交
3376
    BlockBackend *blk = NULL;
3377
    QemuOpts *param;
3378

3379 3380 3381 3382 3383 3384 3385 3386 3387 3388 3389
    static QemuOptsList resize_options = {
        .name = "resize_options",
        .head = QTAILQ_HEAD_INITIALIZER(resize_options.head),
        .desc = {
            {
                .name = BLOCK_OPT_SIZE,
                .type = QEMU_OPT_SIZE,
                .help = "Virtual disk size"
            }, {
                /* end of list */
            }
3390 3391
        },
    };
3392
    bool image_opts = false;
3393

K
Kevin Wolf 已提交
3394 3395 3396
    /* Remove size from argv manually so that negative numbers are not treated
     * as options by getopt. */
    if (argc < 3) {
F
Fam Zheng 已提交
3397
        error_exit("Not enough arguments");
K
Kevin Wolf 已提交
3398 3399 3400 3401 3402 3403
        return 1;
    }

    size = argv[--argc];

    /* Parse getopt arguments */
3404 3405
    fmt = NULL;
    for(;;) {
3406 3407 3408
        static const struct option long_options[] = {
            {"help", no_argument, 0, 'h'},
            {"object", required_argument, 0, OPTION_OBJECT},
3409
            {"image-opts", no_argument, 0, OPTION_IMAGE_OPTS},
3410 3411
            {0, 0, 0, 0}
        };
3412
        c = getopt_long(argc, argv, ":f:hq",
3413
                        long_options, NULL);
3414 3415 3416 3417
        if (c == -1) {
            break;
        }
        switch(c) {
3418 3419 3420
        case ':':
            missing_argument(argv[optind - 1]);
            break;
J
Jes Sorensen 已提交
3421
        case '?':
3422 3423
            unrecognized_option(argv[optind - 1]);
            break;
3424 3425 3426 3427 3428 3429
        case 'h':
            help();
            break;
        case 'f':
            fmt = optarg;
            break;
3430 3431 3432
        case 'q':
            quiet = true;
            break;
3433 3434 3435 3436 3437 3438 3439 3440
        case OPTION_OBJECT: {
            QemuOpts *opts;
            opts = qemu_opts_parse_noisily(&qemu_object_opts,
                                           optarg, true);
            if (!opts) {
                return 1;
            }
        }   break;
3441 3442 3443
        case OPTION_IMAGE_OPTS:
            image_opts = true;
            break;
3444 3445
        }
    }
3446
    if (optind != argc - 1) {
F
Fam Zheng 已提交
3447
        error_exit("Expecting one image file name");
3448 3449 3450
    }
    filename = argv[optind++];

3451 3452
    if (qemu_opts_foreach(&qemu_object_opts,
                          user_creatable_add_opts_foreach,
3453
                          NULL, NULL)) {
3454 3455 3456
        return 1;
    }

3457 3458 3459 3460 3461 3462 3463 3464 3465 3466 3467 3468 3469 3470 3471 3472
    /* Choose grow, shrink, or absolute resize mode */
    switch (size[0]) {
    case '+':
        relative = 1;
        size++;
        break;
    case '-':
        relative = -1;
        size++;
        break;
    default:
        relative = 0;
        break;
    }

    /* Parse size */
3473
    param = qemu_opts_create(&resize_options, NULL, 0, &error_abort);
3474
    qemu_opt_set(param, BLOCK_OPT_SIZE, size, &err);
3475 3476
    if (err) {
        error_report_err(err);
3477
        ret = -1;
3478
        qemu_opts_del(param);
3479
        goto out;
3480
    }
3481 3482
    n = qemu_opt_get_size(param, BLOCK_OPT_SIZE, 0);
    qemu_opts_del(param);
3483

3484
    blk = img_open(image_opts, filename, fmt,
3485 3486
                   BDRV_O_RDWR | BDRV_O_RESIZE, false, quiet,
                   false);
3487
    if (!blk) {
3488 3489
        ret = -1;
        goto out;
3490
    }
3491 3492

    if (relative) {
3493
        total_size = blk_getlength(blk) + n * relative;
3494 3495 3496 3497
    } else {
        total_size = n;
    }
    if (total_size <= 0) {
3498
        error_report("New image size must be positive");
3499 3500
        ret = -1;
        goto out;
3501 3502
    }

3503 3504
    ret = blk_truncate(blk, total_size, &err);
    if (!ret) {
3505
        qprintf(quiet, "Image resized.\n");
3506 3507
    } else {
        error_report_err(err);
3508
    }
3509
out:
M
Markus Armbruster 已提交
3510
    blk_unref(blk);
3511 3512 3513
    if (ret) {
        return 1;
    }
3514 3515 3516
    return 0;
}

3517
static void amend_status_cb(BlockDriverState *bs,
3518 3519
                            int64_t offset, int64_t total_work_size,
                            void *opaque)
3520 3521 3522 3523
{
    qemu_progress_print(100.f * offset / total_work_size, 0);
}

M
Max Reitz 已提交
3524 3525
static int img_amend(int argc, char **argv)
{
3526
    Error *err = NULL;
M
Max Reitz 已提交
3527 3528
    int c, ret = 0;
    char *options = NULL;
3529 3530
    QemuOptsList *create_opts = NULL;
    QemuOpts *opts = NULL;
3531 3532
    const char *fmt = NULL, *filename, *cache;
    int flags;
3533
    bool writethrough;
3534
    bool quiet = false, progress = false;
M
Markus Armbruster 已提交
3535
    BlockBackend *blk = NULL;
M
Max Reitz 已提交
3536
    BlockDriverState *bs = NULL;
3537
    bool image_opts = false;
M
Max Reitz 已提交
3538

3539
    cache = BDRV_DEFAULT_CACHE;
M
Max Reitz 已提交
3540
    for (;;) {
3541 3542 3543
        static const struct option long_options[] = {
            {"help", no_argument, 0, 'h'},
            {"object", required_argument, 0, OPTION_OBJECT},
3544
            {"image-opts", no_argument, 0, OPTION_IMAGE_OPTS},
3545 3546
            {0, 0, 0, 0}
        };
3547
        c = getopt_long(argc, argv, ":ho:f:t:pq",
3548
                        long_options, NULL);
M
Max Reitz 已提交
3549 3550 3551 3552 3553
        if (c == -1) {
            break;
        }

        switch (c) {
3554 3555 3556
        case ':':
            missing_argument(argv[optind - 1]);
            break;
3557
        case '?':
3558 3559 3560
            unrecognized_option(argv[optind - 1]);
            break;
        case 'h':
3561 3562 3563 3564 3565 3566 3567 3568 3569 3570 3571 3572 3573 3574 3575 3576 3577 3578 3579 3580 3581 3582 3583 3584 3585 3586 3587 3588 3589 3590 3591 3592 3593 3594 3595 3596 3597 3598 3599
            help();
            break;
        case 'o':
            if (!is_valid_option_list(optarg)) {
                error_report("Invalid option list: %s", optarg);
                ret = -1;
                goto out_no_progress;
            }
            if (!options) {
                options = g_strdup(optarg);
            } else {
                char *old_options = options;
                options = g_strdup_printf("%s,%s", options, optarg);
                g_free(old_options);
            }
            break;
        case 'f':
            fmt = optarg;
            break;
        case 't':
            cache = optarg;
            break;
        case 'p':
            progress = true;
            break;
        case 'q':
            quiet = true;
            break;
        case OPTION_OBJECT:
            opts = qemu_opts_parse_noisily(&qemu_object_opts,
                                           optarg, true);
            if (!opts) {
                ret = -1;
                goto out_no_progress;
            }
            break;
        case OPTION_IMAGE_OPTS:
            image_opts = true;
            break;
M
Max Reitz 已提交
3600 3601 3602
        }
    }

3603
    if (!options) {
F
Fam Zheng 已提交
3604
        error_exit("Must specify options (-o)");
M
Max Reitz 已提交
3605 3606
    }

3607 3608
    if (qemu_opts_foreach(&qemu_object_opts,
                          user_creatable_add_opts_foreach,
3609
                          NULL, NULL)) {
3610 3611 3612 3613
        ret = -1;
        goto out_no_progress;
    }

3614 3615 3616 3617 3618
    if (quiet) {
        progress = false;
    }
    qemu_progress_init(progress, 1.0);

3619 3620 3621 3622 3623 3624
    filename = (optind == argc - 1) ? argv[argc - 1] : NULL;
    if (fmt && has_help_option(options)) {
        /* If a format is explicitly specified (and possibly no filename is
         * given), print option help here */
        ret = print_block_option_help(filename, fmt);
        goto out;
M
Max Reitz 已提交
3625 3626
    }

3627
    if (optind != argc - 1) {
M
Max Reitz 已提交
3628 3629 3630
        error_report("Expecting one image file name");
        ret = -1;
        goto out;
3631
    }
M
Max Reitz 已提交
3632

3633 3634
    flags = BDRV_O_RDWR;
    ret = bdrv_parse_cache_mode(cache, &flags, &writethrough);
3635 3636 3637 3638 3639
    if (ret < 0) {
        error_report("Invalid cache option: %s", cache);
        goto out;
    }

3640 3641
    blk = img_open(image_opts, filename, fmt, flags, writethrough, quiet,
                   false);
3642
    if (!blk) {
M
Max Reitz 已提交
3643 3644 3645
        ret = -1;
        goto out;
    }
3646
    bs = blk_bs(blk);
M
Max Reitz 已提交
3647 3648 3649

    fmt = bs->drv->format_name;

3650
    if (has_help_option(options)) {
3651
        /* If the format was auto-detected, print option help here */
M
Max Reitz 已提交
3652 3653 3654 3655
        ret = print_block_option_help(filename, fmt);
        goto out;
    }

3656 3657 3658 3659 3660 3661 3662
    if (!bs->drv->create_opts) {
        error_report("Format driver '%s' does not support any options to amend",
                     fmt);
        ret = -1;
        goto out;
    }

C
Chunyan Liu 已提交
3663
    create_opts = qemu_opts_append(create_opts, bs->drv->create_opts);
3664
    opts = qemu_opts_create(create_opts, NULL, 0, &error_abort);
P
Paolo Bonzini 已提交
3665 3666 3667 3668 3669
    qemu_opts_do_parse(opts, options, NULL, &err);
    if (err) {
        error_report_err(err);
        ret = -1;
        goto out;
M
Max Reitz 已提交
3670 3671
    }

3672 3673
    /* In case the driver does not call amend_status_cb() */
    qemu_progress_print(0.f, 0);
3674
    ret = bdrv_amend_options(bs, opts, &amend_status_cb, NULL);
3675
    qemu_progress_print(100.f, 0);
M
Max Reitz 已提交
3676 3677 3678 3679 3680 3681
    if (ret < 0) {
        error_report("Error while amending options: %s", strerror(-ret));
        goto out;
    }

out:
3682 3683
    qemu_progress_end();

3684
out_no_progress:
M
Markus Armbruster 已提交
3685
    blk_unref(blk);
3686 3687
    qemu_opts_del(opts);
    qemu_opts_free(create_opts);
3688 3689
    g_free(options);

M
Max Reitz 已提交
3690 3691 3692 3693 3694 3695
    if (ret) {
        return 1;
    }
    return 0;
}

K
Kevin Wolf 已提交
3696 3697 3698
typedef struct BenchData {
    BlockBackend *blk;
    uint64_t image_size;
K
Kevin Wolf 已提交
3699
    bool write;
K
Kevin Wolf 已提交
3700
    int bufsize;
3701
    int step;
K
Kevin Wolf 已提交
3702 3703
    int nrreq;
    int n;
3704 3705
    int flush_interval;
    bool drain_on_flush;
K
Kevin Wolf 已提交
3706 3707 3708 3709
    uint8_t *buf;
    QEMUIOVector *qiov;

    int in_flight;
3710
    bool in_flush;
K
Kevin Wolf 已提交
3711 3712 3713
    uint64_t offset;
} BenchData;

3714 3715 3716
static void bench_undrained_flush_cb(void *opaque, int ret)
{
    if (ret < 0) {
3717
        error_report("Failed flush request: %s", strerror(-ret));
3718 3719 3720 3721
        exit(EXIT_FAILURE);
    }
}

K
Kevin Wolf 已提交
3722 3723 3724 3725 3726 3727
static void bench_cb(void *opaque, int ret)
{
    BenchData *b = opaque;
    BlockAIOCB *acb;

    if (ret < 0) {
3728
        error_report("Failed request: %s", strerror(-ret));
K
Kevin Wolf 已提交
3729 3730
        exit(EXIT_FAILURE);
    }
3731 3732 3733 3734 3735 3736 3737 3738

    if (b->in_flush) {
        /* Just finished a flush with drained queue: Start next requests */
        assert(b->in_flight == 0);
        b->in_flush = false;
    } else if (b->in_flight > 0) {
        int remaining = b->n - b->in_flight;

K
Kevin Wolf 已提交
3739 3740
        b->n--;
        b->in_flight--;
3741 3742 3743 3744 3745 3746 3747 3748 3749 3750 3751 3752 3753 3754 3755 3756 3757 3758 3759 3760 3761 3762 3763

        /* Time for flush? Drain queue if requested, then flush */
        if (b->flush_interval && remaining % b->flush_interval == 0) {
            if (!b->in_flight || !b->drain_on_flush) {
                BlockCompletionFunc *cb;

                if (b->drain_on_flush) {
                    b->in_flush = true;
                    cb = bench_cb;
                } else {
                    cb = bench_undrained_flush_cb;
                }

                acb = blk_aio_flush(b->blk, cb, b);
                if (!acb) {
                    error_report("Failed to issue flush request");
                    exit(EXIT_FAILURE);
                }
            }
            if (b->drain_on_flush) {
                return;
            }
        }
K
Kevin Wolf 已提交
3764 3765 3766
    }

    while (b->n > b->in_flight && b->in_flight < b->nrreq) {
3767 3768 3769 3770 3771 3772 3773 3774
        int64_t offset = b->offset;
        /* blk_aio_* might look for completed I/Os and kick bench_cb
         * again, so make sure this operation is counted by in_flight
         * and b->offset is ready for the next submission.
         */
        b->in_flight++;
        b->offset += b->step;
        b->offset %= b->image_size;
K
Kevin Wolf 已提交
3775
        if (b->write) {
3776
            acb = blk_aio_pwritev(b->blk, offset, b->qiov, 0, bench_cb, b);
K
Kevin Wolf 已提交
3777
        } else {
3778
            acb = blk_aio_preadv(b->blk, offset, b->qiov, 0, bench_cb, b);
K
Kevin Wolf 已提交
3779
        }
K
Kevin Wolf 已提交
3780 3781 3782 3783 3784 3785 3786 3787 3788 3789 3790 3791 3792
        if (!acb) {
            error_report("Failed to issue request");
            exit(EXIT_FAILURE);
        }
    }
}

static int img_bench(int argc, char **argv)
{
    int c, ret = 0;
    const char *fmt = NULL, *filename;
    bool quiet = false;
    bool image_opts = false;
K
Kevin Wolf 已提交
3793
    bool is_write = false;
K
Kevin Wolf 已提交
3794 3795
    int count = 75000;
    int depth = 64;
3796
    int64_t offset = 0;
K
Kevin Wolf 已提交
3797
    size_t bufsize = 4096;
K
Kevin Wolf 已提交
3798
    int pattern = 0;
3799
    size_t step = 0;
3800 3801
    int flush_interval = 0;
    bool drain_on_flush = true;
K
Kevin Wolf 已提交
3802 3803 3804 3805
    int64_t image_size;
    BlockBackend *blk = NULL;
    BenchData data = {};
    int flags = 0;
3806
    bool writethrough = false;
K
Kevin Wolf 已提交
3807 3808
    struct timeval t1, t2;
    int i;
3809
    bool force_share = false;
K
Kevin Wolf 已提交
3810 3811 3812 3813

    for (;;) {
        static const struct option long_options[] = {
            {"help", no_argument, 0, 'h'},
3814
            {"flush-interval", required_argument, 0, OPTION_FLUSH_INTERVAL},
K
Kevin Wolf 已提交
3815
            {"image-opts", no_argument, 0, OPTION_IMAGE_OPTS},
K
Kevin Wolf 已提交
3816
            {"pattern", required_argument, 0, OPTION_PATTERN},
3817
            {"no-drain", no_argument, 0, OPTION_NO_DRAIN},
3818
            {"force-share", no_argument, 0, 'U'},
K
Kevin Wolf 已提交
3819 3820
            {0, 0, 0, 0}
        };
3821
        c = getopt_long(argc, argv, ":hc:d:f:no:qs:S:t:wU", long_options, NULL);
K
Kevin Wolf 已提交
3822 3823 3824 3825 3826
        if (c == -1) {
            break;
        }

        switch (c) {
3827 3828 3829
        case ':':
            missing_argument(argv[optind - 1]);
            break;
K
Kevin Wolf 已提交
3830
        case '?':
3831 3832 3833
            unrecognized_option(argv[optind - 1]);
            break;
        case 'h':
K
Kevin Wolf 已提交
3834 3835 3836 3837
            help();
            break;
        case 'c':
        {
3838 3839 3840
            unsigned long res;

            if (qemu_strtoul(optarg, NULL, 0, &res) < 0 || res > INT_MAX) {
K
Kevin Wolf 已提交
3841 3842 3843
                error_report("Invalid request count specified");
                return 1;
            }
3844
            count = res;
K
Kevin Wolf 已提交
3845 3846 3847 3848
            break;
        }
        case 'd':
        {
3849 3850 3851
            unsigned long res;

            if (qemu_strtoul(optarg, NULL, 0, &res) < 0 || res > INT_MAX) {
K
Kevin Wolf 已提交
3852 3853 3854
                error_report("Invalid queue depth specified");
                return 1;
            }
3855
            depth = res;
K
Kevin Wolf 已提交
3856 3857 3858 3859 3860 3861 3862 3863
            break;
        }
        case 'f':
            fmt = optarg;
            break;
        case 'n':
            flags |= BDRV_O_NATIVE_AIO;
            break;
3864 3865
        case 'o':
        {
3866 3867
            offset = cvtnum(optarg);
            if (offset < 0) {
3868 3869 3870 3871 3872 3873
                error_report("Invalid offset specified");
                return 1;
            }
            break;
        }
            break;
K
Kevin Wolf 已提交
3874 3875 3876 3877 3878 3879 3880
        case 'q':
            quiet = true;
            break;
        case 's':
        {
            int64_t sval;

3881 3882
            sval = cvtnum(optarg);
            if (sval < 0 || sval > INT_MAX) {
K
Kevin Wolf 已提交
3883 3884 3885 3886 3887 3888 3889
                error_report("Invalid buffer size specified");
                return 1;
            }

            bufsize = sval;
            break;
        }
3890 3891 3892 3893
        case 'S':
        {
            int64_t sval;

3894 3895
            sval = cvtnum(optarg);
            if (sval < 0 || sval > INT_MAX) {
3896 3897 3898 3899 3900 3901 3902
                error_report("Invalid step size specified");
                return 1;
            }

            step = sval;
            break;
        }
K
Kevin Wolf 已提交
3903 3904 3905 3906 3907 3908 3909 3910
        case 't':
            ret = bdrv_parse_cache_mode(optarg, &flags, &writethrough);
            if (ret < 0) {
                error_report("Invalid cache mode");
                ret = -1;
                goto out;
            }
            break;
K
Kevin Wolf 已提交
3911 3912 3913 3914
        case 'w':
            flags |= BDRV_O_RDWR;
            is_write = true;
            break;
3915 3916 3917
        case 'U':
            force_share = true;
            break;
K
Kevin Wolf 已提交
3918 3919
        case OPTION_PATTERN:
        {
3920 3921 3922
            unsigned long res;

            if (qemu_strtoul(optarg, NULL, 0, &res) < 0 || res > 0xff) {
K
Kevin Wolf 已提交
3923 3924 3925
                error_report("Invalid pattern byte specified");
                return 1;
            }
3926
            pattern = res;
K
Kevin Wolf 已提交
3927 3928
            break;
        }
3929 3930
        case OPTION_FLUSH_INTERVAL:
        {
3931 3932 3933
            unsigned long res;

            if (qemu_strtoul(optarg, NULL, 0, &res) < 0 || res > INT_MAX) {
3934 3935 3936
                error_report("Invalid flush interval specified");
                return 1;
            }
3937
            flush_interval = res;
3938 3939 3940 3941 3942
            break;
        }
        case OPTION_NO_DRAIN:
            drain_on_flush = false;
            break;
K
Kevin Wolf 已提交
3943 3944 3945 3946 3947 3948 3949 3950 3951 3952 3953
        case OPTION_IMAGE_OPTS:
            image_opts = true;
            break;
        }
    }

    if (optind != argc - 1) {
        error_exit("Expecting one image file name");
    }
    filename = argv[argc - 1];

3954 3955 3956 3957 3958 3959 3960 3961 3962 3963 3964
    if (!is_write && flush_interval) {
        error_report("--flush-interval is only available in write tests");
        ret = -1;
        goto out;
    }
    if (flush_interval && flush_interval < depth) {
        error_report("Flush interval can't be smaller than depth");
        ret = -1;
        goto out;
    }

3965 3966
    blk = img_open(image_opts, filename, fmt, flags, writethrough, quiet,
                   force_share);
K
Kevin Wolf 已提交
3967 3968 3969 3970 3971 3972 3973 3974 3975 3976 3977 3978
    if (!blk) {
        ret = -1;
        goto out;
    }

    image_size = blk_getlength(blk);
    if (image_size < 0) {
        ret = image_size;
        goto out;
    }

    data = (BenchData) {
3979 3980 3981 3982 3983 3984 3985 3986 3987 3988
        .blk            = blk,
        .image_size     = image_size,
        .bufsize        = bufsize,
        .step           = step ?: bufsize,
        .nrreq          = depth,
        .n              = count,
        .offset         = offset,
        .write          = is_write,
        .flush_interval = flush_interval,
        .drain_on_flush = drain_on_flush,
K
Kevin Wolf 已提交
3989
    };
3990
    printf("Sending %d %s requests, %d bytes each, %d in parallel "
3991
           "(starting at offset %" PRId64 ", step size %d)\n",
3992
           data.n, data.write ? "write" : "read", data.bufsize, data.nrreq,
3993
           data.offset, data.step);
3994 3995 3996
    if (flush_interval) {
        printf("Sending flush every %d requests\n", flush_interval);
    }
K
Kevin Wolf 已提交
3997 3998

    data.buf = blk_blockalign(blk, data.nrreq * data.bufsize);
K
Kevin Wolf 已提交
3999 4000
    memset(data.buf, pattern, data.nrreq * data.bufsize);

K
Kevin Wolf 已提交
4001 4002 4003 4004 4005 4006 4007 4008 4009 4010 4011 4012 4013 4014 4015 4016 4017 4018 4019 4020 4021 4022 4023 4024
    data.qiov = g_new(QEMUIOVector, data.nrreq);
    for (i = 0; i < data.nrreq; i++) {
        qemu_iovec_init(&data.qiov[i], 1);
        qemu_iovec_add(&data.qiov[i],
                       data.buf + i * data.bufsize, data.bufsize);
    }

    gettimeofday(&t1, NULL);
    bench_cb(&data, 0);

    while (data.n > 0) {
        main_loop_wait(false);
    }
    gettimeofday(&t2, NULL);

    printf("Run completed in %3.3f seconds.\n",
           (t2.tv_sec - t1.tv_sec)
           + ((double)(t2.tv_usec - t1.tv_usec) / 1000000));

out:
    qemu_vfree(data.buf);
    blk_unref(blk);

    if (ret) {
R
Reda Sallahi 已提交
4025 4026 4027 4028 4029 4030 4031 4032 4033
        return 1;
    }
    return 0;
}

#define C_BS      01
#define C_COUNT   02
#define C_IF      04
#define C_OF      010
R
Reda Sallahi 已提交
4034
#define C_SKIP    020
R
Reda Sallahi 已提交
4035 4036 4037 4038 4039 4040 4041 4042 4043 4044

struct DdInfo {
    unsigned int flags;
    int64_t count;
};

struct DdIo {
    int bsz;    /* Block size */
    char *filename;
    uint8_t *buf;
R
Reda Sallahi 已提交
4045
    int64_t offset;
R
Reda Sallahi 已提交
4046 4047 4048 4049 4050 4051 4052 4053 4054 4055 4056 4057 4058 4059
};

struct DdOpts {
    const char *name;
    int (*f)(const char *, struct DdIo *, struct DdIo *, struct DdInfo *);
    unsigned int flag;
};

static int img_dd_bs(const char *arg,
                     struct DdIo *in, struct DdIo *out,
                     struct DdInfo *dd)
{
    int64_t res;

4060
    res = cvtnum(arg);
R
Reda Sallahi 已提交
4061

4062
    if (res <= 0 || res > INT_MAX) {
R
Reda Sallahi 已提交
4063 4064 4065 4066 4067 4068 4069 4070 4071 4072 4073 4074
        error_report("invalid number: '%s'", arg);
        return 1;
    }
    in->bsz = out->bsz = res;

    return 0;
}

static int img_dd_count(const char *arg,
                        struct DdIo *in, struct DdIo *out,
                        struct DdInfo *dd)
{
4075
    dd->count = cvtnum(arg);
R
Reda Sallahi 已提交
4076

4077
    if (dd->count < 0) {
R
Reda Sallahi 已提交
4078 4079 4080 4081 4082 4083 4084 4085 4086 4087 4088 4089 4090 4091 4092 4093 4094 4095 4096 4097 4098 4099 4100 4101 4102
        error_report("invalid number: '%s'", arg);
        return 1;
    }

    return 0;
}

static int img_dd_if(const char *arg,
                     struct DdIo *in, struct DdIo *out,
                     struct DdInfo *dd)
{
    in->filename = g_strdup(arg);

    return 0;
}

static int img_dd_of(const char *arg,
                     struct DdIo *in, struct DdIo *out,
                     struct DdInfo *dd)
{
    out->filename = g_strdup(arg);

    return 0;
}

R
Reda Sallahi 已提交
4103 4104 4105 4106
static int img_dd_skip(const char *arg,
                       struct DdIo *in, struct DdIo *out,
                       struct DdInfo *dd)
{
4107
    in->offset = cvtnum(arg);
R
Reda Sallahi 已提交
4108

4109
    if (in->offset < 0) {
R
Reda Sallahi 已提交
4110 4111 4112 4113 4114 4115 4116
        error_report("invalid number: '%s'", arg);
        return 1;
    }

    return 0;
}

R
Reda Sallahi 已提交
4117 4118 4119 4120 4121 4122 4123 4124 4125 4126 4127 4128 4129 4130 4131 4132
static int img_dd(int argc, char **argv)
{
    int ret = 0;
    char *arg = NULL;
    char *tmp;
    BlockDriver *drv = NULL, *proto_drv = NULL;
    BlockBackend *blk1 = NULL, *blk2 = NULL;
    QemuOpts *opts = NULL;
    QemuOptsList *create_opts = NULL;
    Error *local_err = NULL;
    bool image_opts = false;
    int c, i;
    const char *out_fmt = "raw";
    const char *fmt = NULL;
    int64_t size = 0;
    int64_t block_count = 0, out_pos, in_pos;
4133
    bool force_share = false;
R
Reda Sallahi 已提交
4134 4135 4136 4137 4138 4139 4140
    struct DdInfo dd = {
        .flags = 0,
        .count = 0,
    };
    struct DdIo in = {
        .bsz = 512, /* Block size is by default 512 bytes */
        .filename = NULL,
R
Reda Sallahi 已提交
4141 4142
        .buf = NULL,
        .offset = 0
R
Reda Sallahi 已提交
4143 4144 4145 4146
    };
    struct DdIo out = {
        .bsz = 512,
        .filename = NULL,
R
Reda Sallahi 已提交
4147 4148
        .buf = NULL,
        .offset = 0
R
Reda Sallahi 已提交
4149 4150 4151 4152 4153 4154 4155
    };

    const struct DdOpts options[] = {
        { "bs", img_dd_bs, C_BS },
        { "count", img_dd_count, C_COUNT },
        { "if", img_dd_if, C_IF },
        { "of", img_dd_of, C_OF },
R
Reda Sallahi 已提交
4156
        { "skip", img_dd_skip, C_SKIP },
R
Reda Sallahi 已提交
4157 4158 4159 4160 4161
        { NULL, NULL, 0 }
    };
    const struct option long_options[] = {
        { "help", no_argument, 0, 'h'},
        { "image-opts", no_argument, 0, OPTION_IMAGE_OPTS},
4162
        { "force-share", no_argument, 0, 'U'},
R
Reda Sallahi 已提交
4163 4164 4165
        { 0, 0, 0, 0 }
    };

4166
    while ((c = getopt_long(argc, argv, ":hf:O:U", long_options, NULL))) {
R
Reda Sallahi 已提交
4167 4168 4169 4170 4171 4172 4173 4174 4175 4176
        if (c == EOF) {
            break;
        }
        switch (c) {
        case 'O':
            out_fmt = optarg;
            break;
        case 'f':
            fmt = optarg;
            break;
4177 4178 4179
        case ':':
            missing_argument(argv[optind - 1]);
            break;
R
Reda Sallahi 已提交
4180
        case '?':
4181 4182
            unrecognized_option(argv[optind - 1]);
            break;
R
Reda Sallahi 已提交
4183 4184 4185
        case 'h':
            help();
            break;
4186 4187 4188
        case 'U':
            force_share = true;
            break;
R
Reda Sallahi 已提交
4189 4190 4191 4192 4193 4194 4195 4196 4197 4198 4199 4200 4201 4202 4203 4204 4205 4206 4207 4208 4209 4210 4211 4212 4213 4214 4215 4216 4217 4218 4219 4220 4221 4222 4223 4224 4225 4226 4227 4228 4229 4230 4231 4232
        case OPTION_IMAGE_OPTS:
            image_opts = true;
            break;
        }
    }

    for (i = optind; i < argc; i++) {
        int j;
        arg = g_strdup(argv[i]);

        tmp = strchr(arg, '=');
        if (tmp == NULL) {
            error_report("unrecognized operand %s", arg);
            ret = -1;
            goto out;
        }

        *tmp++ = '\0';

        for (j = 0; options[j].name != NULL; j++) {
            if (!strcmp(arg, options[j].name)) {
                break;
            }
        }
        if (options[j].name == NULL) {
            error_report("unrecognized operand %s", arg);
            ret = -1;
            goto out;
        }

        if (options[j].f(tmp, &in, &out, &dd) != 0) {
            ret = -1;
            goto out;
        }
        dd.flags |= options[j].flag;
        g_free(arg);
        arg = NULL;
    }

    if (!(dd.flags & C_IF && dd.flags & C_OF)) {
        error_report("Must specify both input and output files");
        ret = -1;
        goto out;
    }
4233 4234
    blk1 = img_open(image_opts, in.filename, fmt, 0, false, false,
                    force_share);
R
Reda Sallahi 已提交
4235 4236 4237 4238 4239 4240 4241 4242 4243 4244 4245 4246 4247 4248 4249 4250 4251 4252 4253 4254 4255 4256 4257 4258 4259 4260 4261 4262 4263 4264 4265 4266 4267 4268 4269 4270 4271 4272 4273 4274 4275 4276 4277 4278 4279 4280 4281 4282

    if (!blk1) {
        ret = -1;
        goto out;
    }

    drv = bdrv_find_format(out_fmt);
    if (!drv) {
        error_report("Unknown file format");
        ret = -1;
        goto out;
    }
    proto_drv = bdrv_find_protocol(out.filename, true, &local_err);

    if (!proto_drv) {
        error_report_err(local_err);
        ret = -1;
        goto out;
    }
    if (!drv->create_opts) {
        error_report("Format driver '%s' does not support image creation",
                     drv->format_name);
        ret = -1;
        goto out;
    }
    if (!proto_drv->create_opts) {
        error_report("Protocol driver '%s' does not support image creation",
                     proto_drv->format_name);
        ret = -1;
        goto out;
    }
    create_opts = qemu_opts_append(create_opts, drv->create_opts);
    create_opts = qemu_opts_append(create_opts, proto_drv->create_opts);

    opts = qemu_opts_create(create_opts, NULL, 0, &error_abort);

    size = blk_getlength(blk1);
    if (size < 0) {
        error_report("Failed to get size for '%s'", in.filename);
        ret = -1;
        goto out;
    }

    if (dd.flags & C_COUNT && dd.count <= INT64_MAX / in.bsz &&
        dd.count * in.bsz < size) {
        size = dd.count * in.bsz;
    }

R
Reda Sallahi 已提交
4283 4284 4285 4286 4287 4288 4289 4290
    /* Overflow means the specified offset is beyond input image's size */
    if (dd.flags & C_SKIP && (in.offset > INT64_MAX / in.bsz ||
                              size < in.bsz * in.offset)) {
        qemu_opt_set_number(opts, BLOCK_OPT_SIZE, 0, &error_abort);
    } else {
        qemu_opt_set_number(opts, BLOCK_OPT_SIZE,
                            size - in.bsz * in.offset, &error_abort);
    }
R
Reda Sallahi 已提交
4291 4292 4293 4294 4295 4296 4297 4298 4299 4300 4301

    ret = bdrv_create(drv, out.filename, opts, &local_err);
    if (ret < 0) {
        error_reportf_err(local_err,
                          "%s: error while creating output image: ",
                          out.filename);
        ret = -1;
        goto out;
    }

    blk2 = img_open(image_opts, out.filename, out_fmt, BDRV_O_RDWR,
4302
                    false, false, false);
R
Reda Sallahi 已提交
4303 4304 4305 4306 4307 4308

    if (!blk2) {
        ret = -1;
        goto out;
    }

R
Reda Sallahi 已提交
4309 4310 4311 4312 4313 4314 4315 4316 4317 4318 4319
    if (dd.flags & C_SKIP && (in.offset > INT64_MAX / in.bsz ||
                              size < in.offset * in.bsz)) {
        /* We give a warning if the skip option is bigger than the input
         * size and create an empty output disk image (i.e. like dd(1)).
         */
        error_report("%s: cannot skip to specified offset", in.filename);
        in_pos = size;
    } else {
        in_pos = in.offset * in.bsz;
    }

R
Reda Sallahi 已提交
4320 4321
    in.buf = g_new(uint8_t, in.bsz);

R
Reda Sallahi 已提交
4322
    for (out_pos = 0; in_pos < size; block_count++) {
R
Reda Sallahi 已提交
4323 4324 4325 4326 4327 4328 4329 4330 4331 4332 4333 4334 4335 4336 4337 4338 4339 4340 4341 4342 4343 4344 4345 4346 4347 4348 4349 4350 4351 4352 4353 4354 4355 4356 4357 4358 4359 4360
        int in_ret, out_ret;

        if (in_pos + in.bsz > size) {
            in_ret = blk_pread(blk1, in_pos, in.buf, size - in_pos);
        } else {
            in_ret = blk_pread(blk1, in_pos, in.buf, in.bsz);
        }
        if (in_ret < 0) {
            error_report("error while reading from input image file: %s",
                         strerror(-in_ret));
            ret = -1;
            goto out;
        }
        in_pos += in_ret;

        out_ret = blk_pwrite(blk2, out_pos, in.buf, in_ret, 0);

        if (out_ret < 0) {
            error_report("error while writing to output image file: %s",
                         strerror(-out_ret));
            ret = -1;
            goto out;
        }
        out_pos += out_ret;
    }

out:
    g_free(arg);
    qemu_opts_del(opts);
    qemu_opts_free(create_opts);
    blk_unref(blk1);
    blk_unref(blk2);
    g_free(in.filename);
    g_free(out.filename);
    g_free(in.buf);
    g_free(out.buf);

    if (ret) {
K
Kevin Wolf 已提交
4361 4362 4363 4364 4365 4366
        return 1;
    }
    return 0;
}


A
Anthony Liguori 已提交
4367
static const img_cmd_t img_cmds[] = {
4368 4369 4370 4371 4372 4373 4374 4375
#define DEF(option, callback, arg_string)        \
    { option, callback },
#include "qemu-img-cmds.h"
#undef DEF
#undef GEN_DOCS
    { NULL, NULL, },
};

B
bellard 已提交
4376 4377
int main(int argc, char **argv)
{
A
Anthony Liguori 已提交
4378
    const img_cmd_t *cmd;
4379
    const char *cmdname;
4380
    Error *local_error = NULL;
4381
    char *trace_file = NULL;
4382 4383 4384
    int c;
    static const struct option long_options[] = {
        {"help", no_argument, 0, 'h'},
4385
        {"version", no_argument, 0, 'V'},
4386
        {"trace", required_argument, NULL, 'T'},
4387 4388
        {0, 0, 0, 0}
    };
B
bellard 已提交
4389

4390 4391 4392 4393
#ifdef CONFIG_POSIX
    signal(SIGPIPE, SIG_IGN);
#endif

4394
    module_call_init(MODULE_INIT_TRACE);
K
Kevin Wolf 已提交
4395
    error_set_progname(argv[0]);
4396
    qemu_init_exec_dir(argv[0]);
K
Kevin Wolf 已提交
4397

4398
    if (qemu_init_main_loop(&local_error)) {
4399
        error_report_err(local_error);
4400 4401 4402
        exit(EXIT_FAILURE);
    }

4403
    qcrypto_init(&error_fatal);
4404

4405
    module_call_init(MODULE_INIT_QOM);
B
bellard 已提交
4406
    bdrv_init();
F
Fam Zheng 已提交
4407 4408 4409
    if (argc < 2) {
        error_exit("Not enough arguments");
    }
4410

4411
    qemu_add_opts(&qemu_object_opts);
4412
    qemu_add_opts(&qemu_source_opts);
4413
    qemu_add_opts(&qemu_trace_opts);
4414

4415
    while ((c = getopt_long(argc, argv, "+:hVT:", long_options, NULL)) != -1) {
4416
        switch (c) {
4417 4418 4419
        case ':':
            missing_argument(argv[optind - 1]);
            return 0;
4420
        case '?':
4421 4422 4423
            unrecognized_option(argv[optind - 1]);
            return 0;
        case 'h':
4424 4425 4426 4427 4428
            help();
            return 0;
        case 'V':
            printf(QEMU_IMG_VERSION);
            return 0;
4429 4430 4431 4432
        case 'T':
            g_free(trace_file);
            trace_file = trace_opt_parse(optarg);
            break;
4433
        }
B
bellard 已提交
4434
    }
4435

4436
    cmdname = argv[optind];
4437

4438 4439 4440
    /* reset getopt_long scanning */
    argc -= optind;
    if (argc < 1) {
4441 4442
        return 0;
    }
4443
    argv += optind;
D
Denis V. Lunev 已提交
4444
    optind = 0;
4445

4446 4447 4448 4449 4450 4451
    if (!trace_init_backends()) {
        exit(1);
    }
    trace_init_file(trace_file);
    qemu_set_log(LOG_TRACE);

4452 4453 4454 4455 4456 4457
    /* find the command */
    for (cmd = img_cmds; cmd->name != NULL; cmd++) {
        if (!strcmp(cmdname, cmd->name)) {
            return cmd->handler(argc, argv);
        }
    }
4458

4459
    /* not found */
F
Fam Zheng 已提交
4460
    error_exit("Command not found: %s", cmdname);
B
bellard 已提交
4461
}